HST

m B

:M'M''

H

t I

THE

VOYAGE OF H.M.S. CHALLENGER.

ZOOLOGY-VOL. IV.

REPORT

ON THE

She

SCIENTIFIC RESULTS

OF THE

VOYAGE OF H.M.S. CHALLENGER

DURING THE YEARS i 8 7 3-7 6

UNDER THE COMMAND OF

Captain GEORGE S. NARES, R.N., F.R.S.

AND

Captain FRANK TOURLE THOMSON, R.N.

PREPARED UNDER THE SUPERINTENDENCE OF

THE LATE

Sir C. WYVILLE THOMSON, Knt., F.R.S., &c.

REGIUS PROFESSOR OF NATURAL HISTORY IN THE UNIVERSITY OF EDINBURGH
DIRECTOR OF THE CIVILIAN SCIENTIFIC STAFF ON BOARD

AND NOW OF

JOHN MURRAY, F.R.S.E.

ONE OF THE NATURALISTS OF THE EXPEDITION

Zoology — Vol. IV.

IPubltsJjeD bp <0rOer of l£>er jflajestp's ©obernment

PRINTED FOR HER MAJESTY'S STATIONERY OFFICE

AND SOLD BY

LONDON:— LONGMANS & CO. ; JOHN MURRAY; MACMILLAN & CO.; SIMPKIN, MARSHALL, & CO.
TRUBNER & CO. ; E. STANFORD ; J. U. POTTER ; AND KEGAN PAUL, TRENCH, & CO.
EDINBURGH :— ADAM & CHARLES BLACK and DOUGLAS & FOULIS.
DUBLIN :— A. THOM & CO. and HODGES, FIGGIS, & CO.

l8S2

Price Fifty Shillings

PRINTED BY NEILL AND COMPANY, EDINBURGH,
FOR HER MAJESTY'S STATIONERY OFFICE.

THE LATE SIR C. WYVILLE THOMSON.

The announcement of the death of Sir Charles Wyville Thomson, since
the last volume of these Reports was issued from the press, has been received
with general regret.

Sir Wyville's name will always be associated with deep-sea investiga-
tions, and in a very special manner with those conducted by H.M.S.
Challenger. He was one of the chief promoters of the Expedition, and Avas on
board the ship during the whole of her long cruise as Director of the Civilian
Scientific Staff. Until the beginning of the present year he had the direc-
tion of all the work connected with the publication of these Official Eeports.

A subsequent volume will contain a history of the Challenger Expedition,
and of those English Expeditions which immediately preceded it, as well as
a detailed account of Sir Wyville's scientific work in connection therewith.
The present notice is limited to an enumeration of some of the chief events
of his life.

Charles Wyville Thomson, the son of Mr. Andrew Thomson, surgeon
in the Honourable East India Company's service, was born at Bonsyde, near
Linlithgow, on 5th March 1830.

He received his earlv education at Merchiston Castle School, and after-
wards attended the classes in the University of Edinburgh as a student of
medicine.

In 1850 he became Lecturer on Botany in King's College, Aberdeen ; and
in 1851 he was appointed to the Chair of Botany in Marischal College.

vi THE VOYAGE OF H.M.S. CHALLENGER.

In 1853 he was appointed Professor of Natural History in Queen's College,
Cork; in 1854 he became Professor of Mineralogy and Geology in Queen's
College, Belfast, and afterwards, in 1860, Professor of Natural Science in the
same college; in 1868 he held the additional post of Professor of Botany in
the Royal College of Science, Dublin.

He was associated with Dr. W. B. Carpenter, C.B., F.R.S., and Dr. J.
Gwyn Jeffreys, F.R.S., in the deep-sea explorations in the North Atlantic,
conducted by H.M.SS. "Porcupine" and "Lightning" in 1868 and 1869.

In 1870 he was appointed Professor of Natural History in the University
of Edinburgh.

In 1872 he obtained leave of absence from the Senatus Academicus of
Edinburgh University, on his appointment by the Lords Commissioners of
the Admiralty, as Director of the Civilian Scientific Staff on board H.M.S.
Challenger.

On the return of the Challenger to England in 1876, he received the
honour of knighthood, and was appointed by the Lords Commissioners of Her
Majesty's Treasury, Director of the " Challenger Expedition Commission."
In the same year he was awarded a Royal medal by the Royal Society
for his successful direction of the scientific investigations carried on by H.M.S.
Challenger.

"&v

In 1877 he was one of the deputation from Edinburgh University to
Sweden, on the occasion of the quatercentenary of the foundation of the
University of Upsala, and on that occasion was created, by King Oscar, a
knioht of the Order of the Polar Star.

'B'

In 1877 he delivered the Rede Lecture at Cambridge, and in 1878 he was
President of the Geographical Section of the British Association at Dublin.

Sir Wyville had a severe illness in the summer of 1879 ; he continued after
that date in failing health, and was not again able to deliver his lectures in
the University of Edinburgh. He died at his residence, Bonsyde, near Lin-
lithgow, on the 10th of March 1882.

THE LATE SIR C. WYVILLE THOMSON. vii

Sir C. Wyville Thomson was an LL.D. of Aberdeen (1853); F.R.S.E.
(1855); an Honorary LL.D. of the Queen's University, Ireland (I860); Fellow
of the Royal Irish Academy (1861); F.R.S. London (1869); Honorary D.Sc.,
Queen's University, Ireland (1871); LL.D. Dublin (1878) ; Ph.D. Jena; and
a Fellow of the Linnean, Geological, Zoological, and Palseontological Societies
of London, and of various Foreign and Colonial Institutes.

The following is a list of Sir Wyville Thomson's principal publications : —

On the Application of Photography to the Comp. Microscope, Brit. Assoc.
Rep., Part 2, 1850.

Notes on some Scotch Zoophytes and Polyzoa, Ann. Nat. Hist., ix., 1852.

On the Character of the Sertularian Zoophytes, Brit. Assoc. Rep., Part 2, 1852.

Notes on some British Zoophytes, Ann. Nat. Hist., xi., 1853.

On some of the Native Irish Zoophytes, Nat. Hist. Rev., ii., 1855.

On the Embryogeny of Comatula rosacea, Linck., Roy. Soc. Proc, ix., 1857-59.

On some Species of Acidasp>is from Silurian Beds of South of Scotland, Quart.
Jour Geol. Soc, 1857.

Description of Loricula macadami, a new fossil Cirripede, Ann. and Mag., 1858.

On the Genus and Species of a new Polyzoon, in the collection of Professor
Harvey, Dublin Zool. Bot. Assoc. Proc, i., 1859.

On a New Palaeozoic Group of the Echinodermata, Edin. Neiu. Phil. Jour.,
xiii., 1861.

On the Embryology of Asteracanthion violaceus, Lin., Microscopic Soc. Jour., i.,
1861-62.

On the Development of Synapta inhcerens, Microscopic Soc. Jour., ii., 1862.

viii THE VOYAGE OF H.M.S. CHALLENGER.

On Distorted Human Skulls, Nat. Hist. Review, 1862.

On the Embryology of Echinodermata, Nat. Hist. Review, 1863.

Do. do. do. 1864.

On Professor Steenstrup's " Views on the Obliquity of Flounders," Ann. Mag.
Nat. Hut, 1865.

Sea Lilies (Cmocrinus — Neocrinus — -Comatuki), Intellect, Obs., vi., 1865.

On the Glass Rope (Hyalonema), Intellect, Obs., xi., 1867.

On the Vitreous Sponges, Ann. Mag. Nat. Hist., 1868.

On Holtenia, a Genus of Vitreous Sponges, Phil. Trans., clix., 1869.

On the Depths of the Sea, Roy. Dublin Soc. Jour., v., 1870.

Osteology of Polypterus, Jour. Anat. and Rhys., 1870.

On Deep-Sea Climates, Nature, ii., 1870.

Preliminary Report, in connection with Drs. W. B. Carpenter and J. G. Jeffreys,
of the Scientific Exploration of the Deep-Sea in H.M.S. " Porcupine,"
Roy. Soc. Proc, xviii., 1870.

On the Distribution of Temperature in the North Atlantic, Nature, iv., 1871.

On the Continuity of the Chalk, Nature, iii., 1871.

On the Structure of The Palaeozoic Crinoids, Edin. Roy. Soc. Proc, vii.,
1871.

Notice of a New Family (Echinothuridge) of the Echinodermata, Edin. Roy.
Soc. Proc, vii., 1872.

Deep-Sea Echinidea, Ann. Mag. Nat. Hist., 1872.

On the Crinoidea of the "Porcupine," Edin. Roy. Soc. Proc, vii., 1872.

Opening Address on the Ripening and Decay of Fruit, Edin. Bot. Soc. Trans.,
1873.

THE LATE SIR C. WYVILLE THOMSON. IX

On the Echinidea of the "Porcupine," Phil. Trans., 1874; Proc. Roy. Soc,

1872.

The Depths of the Sea (1 vol.), 1873. Macmillan & Co. 4to.

On Dredgings and Deep-Sea Soundings in the South Atlantic, in a letter to
Admiral Richards, C.B., F.R.S., Roy. Soc. Proc, 1874.

Preliminary Notes on the Nature of the Sea-Bottom, procured by the Soundings
of H.M.S. Challenger during her cruise in the Southern Sea in 1874, Roy.
Soc. Proc, 1874.

Report to the Admiralty on the Cruise of H.M.S. Challenger from July to
November 1874, Roy. Soc. Proc, 1875.

Report to the Admiralty on the Cruise of H.M.S. Challenger from June to
August 1875, Roy. Soc Proc, 1875.

Preliminary Report to Admiralty on Cruise of the Challenger between Hawaii
and Valparaiso, Proc. Roy. Soc, 1876.

Preliminary Report to Admiralty on Cruise of the Challenger from Falkland
Island to Monte Video, Proc Roy. Soc, 1876.

On the Structure and Relations of the Genus Holopus, Edin. Roy. Soc Proc.

1877.

The Atlantic (2 vols.), 1877. Macmillan & Co. 4to.

Some Peculiarities in the Mode of Propagation of certain Echinoderms of the
Southern Sea, Linn. Soc Joum. Zoology, vol. xiii., 1878.

The General Introduction to the Zoological Series of Reports of the Voyage
of the Challenger, vol. i., Zool., 1880.

CONTENTS.

I. — Report on the Anatomy of the Petrels (Tubinares) collected during the
Voyage of H.M.S. Challenger, in the years 1873-1876.

By W. A. Forbes, B.A., F.L.S., F.G.S., M.B.O.U., Fellow of St. John's College,
Cambridge, Prosector to the Zoological Society of London.

(Received May 6, 1882.)

II. — Report on the Deep-Sea Medusa dredged by H.M.S. Challenger,

during the years 1873-1876.

By Professor Ernst Haeckel, M.D. , Ph.D.
(Received March 1, 1881.)

III. — Report on the Holothurioidea, dredged by H.M.S. Challenger during

the years 1873-1876. Part I.

By Hjalmar Theel.
(Received June 1, 1881.)

79077

EDITOEIAL NOTES.

After the lamented death of Professor A. H. Garrod, Mr. W. A. Forbes
was induced to undertake the anatomical examination of the Petrels collected
during the cruise of the Challenger. The result of Mr. Forbes' labours is
given in the first Report in this volume. This contribution will be found a
most valuable addition to the literature on this remarkable order of Pelagic
birds.

The name which has been given to Professor Haeckel's beautifully illus-
trated Memoir may perhaps be open to question. Some of the species
described are certainly from the surface, and it is even doubtful if all the
others are to be regarded as true deep-sea animals.

Professor Haeckel discusses this question in his Preface, and I quite agree
with him in regarding some of the forms as constant inhabitants of the
deep sea.

In Dr. Theel's Memoir we have the description of a new order of
characteristic deep-sea animals of very great interest.

The majority of the Elasipoda live in depths greater than 1000 fathoms,
and up to the present time only one species has been found in depths
less than 100 fathoms. This species was dredged in the Arctic Ocean, during
the Swedish Expedition to the Yenisei in the year 1875, and was described
by Dr. Th^el shortly after the return of the Challenger to England. This
circumstance led to the whole of the Holothurioidea of the Challenger
Expedition being placed in Dr. Theel's hands for description. Dr. Theel
promises the second part of his very valuable contribution in the course of
a few months.

Owing to the continued ill-health of the late Sir C. Wyville Thomson,
I was, in January of the present year, instructed to undertake the Editorial
duties connected with the Official Publications on the Scientific Results of
the Challenger Expedition ; and, at the same time was requested to prepare.

xiv THE VOYAGE OF H.M.S. CHALLENGER.

for the information of H.M. Government, a Report on the state of the work
in connection therewith.

The List which is here annexed shows the Memoirs already published,
and those in progress or about to be undertaken on the 1st of April 1882 —
the commencement of the current financial year.

JOHN MURRAY.

PUBLISHED MEMOIRS.

General Introduction to the Zoological Series of Eeports.
By Sir C. Wyville Thomson, Knt, F.R.S., &c.

Report on the Brachiopoda.

By Thomas Davidson, F.R.S., F.L.S., F.G.S., V.-P.P.S., &c.

Report on the Pennatulida.

By Professor Albert V. Kolliker, F.M.R.S., &c.

Report on the Ostracoda.

By G. Stewardson Brady, M.D., F.L.S.

Report on the Bones of Cetacea.

By Professor William Turner, M.B. (Lond.), F.R.SS. L. & E.

Report on the Development of the Green Turtle.

By William Kitchen Parker, F.R.S., F.L.S., F.Z.S.

Report on the Shore Fishes.

By Albert Gunther, M.A., M.D., Ph.D, F.R.S., &c.

Report on the Corals.

By Professor H. N. Moseley, M.A., F.R.S.

Report on the Birds.

a. On the Birds collected in the Philippine Islands.

By Arthur, Marquis of Tweeddale, F.R.S.S. L. & K, &c.

EDITORIAL NOTES. XV

/9. On the Birds collected in the Admiralty Islands.
By P. L. Sclater, M.A., PLD., F.E.S.

7. On the Birds collected in Tongatabu, the Fiji Islands, Api (New Hebrides), and Tahiti.

By 0. Finsch, C.M.Z.S.

8. On the Birds collected in Ternate, Amboyna, Banda, the Ki Islands, and the Arrou

Islands.
By T. Salvadori, C.M.Z.S.

e. On the Birds collected at Cape York, Australia, and on the Neighbouring Islands
(Raine, Wednesday, and Booby Islands).
By W. A. Forbes, B.A., F.L.S., F.G.S., F.Z.S.

£ On the Birds collected in the Sandwich Islands.
By P. L. Sclater, M.A., Ph.D., F.E.S., F.L.S.

7). On the Birds collected in Antarctic America.

By P. L. Sclater, MA., Ph.D., F.R.S., and Osbert Salvin, M.A., F.E.S.

6. On the Birds collected on the Atlantic Islands, and Kerguelen Island, and on the
Miscellaneous Collections.
By P. L. Sclater, M.A., PLD., F.E.S.

t. On the Steganopodes and Impennes.

By P. L. Sclater, M.A., PLD., F.R.S., and Osbert Salvin, M.A., F.E.S., F.L.S.

k. On the Larid^e.

By Howard Saunders, F.E.S., F.L.S.

X. On the ProcellaridvE.

By Osbert Salvin, M.A., F.E.S., &c.

(i. List of the Eggs collected.

By P. L. Sclater, M.A., PLD., F.E.S., F.L.S.

v. Note on the Gizzard and other Organs of Carpophaga latrans.
By A. H. Garrod, M.A., F.E.S.

Eeport on the Echinoidea.
By Alexander Agassiz.

Eeport on the Pycnogonida.
By P. P. C. Hoek.

xvi THE VOYAGE OF H.M.S. CHALLENGER.

MEMOIRS TO APPEAR IN THE FINANCIAL YEAR 1882-83.

Report on the Anatomy of the Petrels.

By W. A. Forbes, B.A., F.L.S., F.G.S., F.Z.S.

Report on the Holothurioidea. Part I.
By Hjalmar Theel.

Report on the Deep-Sea Medusa.

By Professor Ernst Haeckel, M.D., Ph.D.

Report on the Ophiuroidea.
By Theodore Lyman.

Report on the Actiniaria.

By Professor Richard Hertwig.

Report on the Petrology of St. Paul's Rocks.
By Professor A. Renard, F.G.S.

Report on the Pressure Errors of the Challenger Thermometers.
By Professor P. G. Tait, M.A., Sec. R.S.E.

Magnetical Results.

By Commander Maclear, R.N. ; Lieutenant Bromley, R.N. ; Staff-Commander Tizard, R.N. ;
and Staff-Commander E. W. Creak, R.N. ; with Instructions and Memorandum prepared
under the Superintendence of the Hydrographer of the Admiralty.

Meteorological Observations.

By Staff-Commander Tizard, R.N., assisted by other Officers of the Ship.

Report on the Polyzoa. Part I.

By George Busk, F.R.S., V.-P.L.S., F.G.S., F.R.C.S.

Report on the Hydroida. Part I.

By Professor G. J. Allman, M.D., LL.D., F.R.SS. L. and E., M.R.I.A., V.-P.L.S.

Report on the Anatomy of the Spheniscid^e. Part I.
By Professor Morrison Watson, M.D., F.R.S.E.

Report on the Marsupialia.

By Professor D. J. Cunningham, M.D., F.R.S.E., F.R.C.S.

EDITORIAL NOTES. xvii

Report on the Copepoda.

By G. Stewaedson Brady, M.D., F.R.S., F.L.S.

Report on the Nudibranchiata.
By Rudolph Bergh.

Chemical Analyses.

By Professor W. Dittmar, F.R.SS. L. and E.

Report on Human Crania.

By Professor William Turner, M.B. (London), F.R.SS. L. and E.

Specific Gravity Observations.

By J. Y. Buchanan, M.A., F.R.S.E.

Report on Diatomace^e.

By Conte Abate Francesco Castracane.

Report on the Tunicata. Part I.

By Professor W. A. Herdman, D.Sc, F.R.S.E.

Report on the Genus Orbitolites.

By W. B. Carpenter, C.B., M.D., LLC, F.R.S., F.G.S., V.-P.L.S., &c.

Report on the Genus Halobates.

By F. Buchanan White, M.D., F.L.S.

MEMOIRS TO FOLLOW IN SUBSEQUENT YEARS.

Report on the Holothurioidea. Part II.
By Hjalmar Theel.

Report on the Polyzoa. Part II.

By George Busk, F.R.S., V.P.L.S., F.R.C.S., F.G.S.

Report on the Cirripedia.
By P. P. C. Hoek.

Report on the Foraminifera.

By H. B. Brady, F.R.S., F.L.S., F.G.S.

Report on the Radiolaria.

By Professor Ernst Haeckel, M.D., Ph.D.
c

xviii THE VOYAGE OF H.M.S. CHALLENGER.

The Narrative of the Voyage. (Vol. I.)

By Staff-Coinmander Tizard, R.N., Professor H. N. Moseley, F.R.S., J. Y. Buchanan, M.A.,
F.R.S.E., and John Murray, F.R.S.E.

Report on the Myzostomid^e.
By Ludwig Graff.

Report on the Alcyonaria.

By Professor E. Perceval Wright, M.D., F.L.S.

Report on the Macrura.

By C. Spence Bate, F.R.S., F.L.S.

The Botany of the Expedition.

By Sir Joseph Dalton Hooker, K.C.S.I., M.D., D.C.L., LL.D., F.R.S., V.-P.L.S., F.G.S , &c.

Report on Marine Mammalia.

By Professor William Turner, M.B. (Lond.), F.R.SS. L. & E.

Report on the Cephalopoda.

By Professor Thomas Henry Huxley, V.-P.R.S., LL.D., &c.

Report on the Gephyrea.

By Professor E. Ray Lankester, M.A., F.R.S., F.L.S.

Report on the Tunicata. Part II.

By Professor W. A. Herdman, D.Sc, F.R.S.E.

Report on the Crinoidea.

By Sir C. Wyville Thomson, Knt., F.R.SS. L. and E., and P. H. Carpenter, M.A.

Report on the Asteroidea.

By W. P. Sladen, F.L.S., F.G.S.

Report on the Annelida

By W. C. M'Intosh, M.D., LL.D., F.R.SS. L. and E., F.L.S., &c.

Report on the Cumacea and Schizopoda.
By

Report on the Anatomy of the SphenisciDjE. Part II.
By Professor Morrison Watson, M.D., F.R.S.E.

EDITORIAL NOTES. xix

Report on the Gasteropoda.

By the Rev. R. Boog Watson, F.R.S.E., F.L.S., and the Marquis de Folin.

Report on the Hydroida. Part II.

By Professor G. J. Allman, M.D., LL.D., F.R.SS. L. and E„ M.R.I.A., V.-P.L.S.

Report on the Deep-Sea Fishes.

By Albert Gunther, M.A., M.D., Ph.D., F.R.S., &c.

Report on the Comatulid^e.

By P. H. Carpenter, M.A.

Report on Deep-Sea Deposits.

By John Murray, F.R.S.E., and Professor A. Renard, F.G.S.

Report on the Hexactinellid^e.
By Professor F. E. Schulze.

Report on the Lamellibranchiata.
By E. A. Smith.

The Anatomy of the Deep-Sea Mollusca.
By

Ocean Circulation.

By Professor P. G. Tait, M.A., Sec. R.S.E., and Alexander Buchan, M.A, F.R.S.E.

The Anatomy of Peripatus.

By Professor F. M. Balfour, M.A, F.R.S.

Report on the MonactinellidjE.
By S. 0. Ridley, F.L.S.

Report on the Tetractinellid^e.

By Professor W. J. Sollas, M.A., F.R.S.E.

Report on the Brachyura.
By E. J. Miers, F.L.S.

Report on the Amphipoda.

By the Rev. T. R. R. Stebbing.

Report on the Stomatopoda.
By

XX THE VOYAGE OF H.M.S. CHALLENGES.

Eepokt on the Anomuka.

By Jules Barrois, Director of the Zoological Laboratory at Villefranche.

Eepokt on the Pteropoda and Heteropoda.
By A. E. Craven, F.L.S., F.G.S.

Concluding Part with Index.
By John Murray, F.R.S.E.

The Isopoda, and one or two other Groups still remain to be allotted.

THE

VOYAGE OF H.M.S. CHALLENGER

ZOOLOGY.

REPORT on the Anatomy of the Petrels (lubinares), collected during the
Voyage of H.M.S. Challenger. By W. A. Forbes, B.A., F.L.S., F.G.S.,
M.B.O.U., Fellow of St. John's College, Cambridge, Prosector to the
Zoological Society of London.

I. INTRODUCTORY.

Materials for the knowledge of the structure of the soft parts of the class Aves, when
the members of that group indigenous to a country have been examined, are for the
most part only to be obtained through the medium of zoological gardens, by the inhabi-
tants of these, on their decease, coming into the hands of some person competent to
examine them. In spite of the increased facilities of communication of the present day,
and the greater experience of those in charge of bving zoological collections, there still
remain many groups of birds which as yet it has been found impossible to obtain or keep
in a living state. Such birds in consequence can only be adequately studied from spirit-
specimens, and these also it is frequently very difficult to obtain, especially if the species
wanted are of large size, or inhabit little explored and inaccessible countries.

The group of Petrels is one that has till the present been hardly at all examined
anatomically, as but few species inhabit the European seas, and even these, on account of
their peculiar habits, are rarely to be obtained in the flesh, either in a living or dead
state. The majority of the group, inhabiting the little visited oceans and islands of the
Southern Hemisphere, have been known simply from skins or skeletons, the great size of
many of them rendering bringing their bodies home in spirit impracticable to any
ordinary collector. Nor have we as yet succeeded in obtaining or keeping any in a living
state, except on one or two rare occasions.

When therefore H.M.S. Challenger was starting on her voyage of circumnavigation it

V (ZOOL. CHALL. EXP. — PART XI. — 1882.) L 1

2 THE VOYAGE OF H.M.S. CHALLENGER.

seemed that an excellent opportunity would be afforded for obtaining material to fill up
the blank that thus existed in our knowledge of the Petrels. At the suggestion of my
lamented predecessor Prof. A. H. Garrod, at that time Prosector to the Zoological Society,
the naturalist staff was requested to pay special attention to forming a collection of these
oceanic birds in spirit, so as to be available for anatomical examination. The result was
a very considerable collection indeed of the birds in question, all excellently preserved,
and including nearly all the most important and interesting of the known genera. These
were handed over, when the collections were being broken up for working out, to
Professor Garrod for examination. Unfortunately he had hardly commenced to work
seriously on them before he was struck down by the fingering illness which eventually
proved fatal to him. During that time, whenever well enough to do so, he continued to
work away at his favourite subject, and many of his drawings made then, chiefly relating
to the conformation of the syrinx in these birds, are now before me. An unfinished MS.
paper of his written about that time, treating on the anatomy of the Diving Petrel
(PelecancM.es) — a form the Procellarian affinities of which were then doubtful — was
sufficiently complete and important to justify, in the writer's opinion, its publication in
the reprint of Professor Garrod's papers which has since been edited by him.

Succeeding to Professor Garrod's position at the Zoological Gardens early in 1880, I
applied immediately to the late Professor Sir Wyville Thomson to be allowed to retain
so much of the material collected by the Challenger as was likely to prove of service to
me in my researches on the anatomy of birds, and I especially asked to be allowed to
retain the collection of Petrels, with the object of drawing up a report thereon for the
present series of papers. I must take this opportunity to record my best thanks to
Sir Wyville Thomson for the very ready way in which he acceded to both my requests.
Having commenced work on the specimens of Petrels collected by the Challenger it
seemed desirable to make my report on the structure of that group as perfect and com-
plete as possible, and during the past two years I have therefore taken every opportunity
of acquiring specimens of them fit for dissection. By these means I have been enabled
to examine several species and genera of these birds not represented in the Challenger
collection, though that collection has formed the groundwork of my investigations.

I herewith give a complete list of those species that I have been enabled to examine
in the flesh. All not otherwise indicated were collected by H.M.S. Challenger. And I
must take this opportunity to thank my friend Mr. Osbert Salvin, F.E.S., who reported
on the collection of Tubinares made in skins during the voyage,1 for his kind assistance in
naming the spirit-specimens under my charge, as well as for much subsequent assistance
in points of nomenclature, and for valuable material that would not have otherwise been
available.

1 Zoology of the Voyage of H.M.S. Challenger, vol. ii. part viii. pp. 140-149 (Report on the Birds — XL On
the Procellariul« collected during the Expedition). Also Proc. Zool. Soc, 1878, pp. 735-740.

report on the anatomy of the petrels.
List of Material Examined.

Number

Name of Bird.

of
Specimens.

Eemarks.

OCEANITID.E

Oceaniles oceanicus, ....

3

One from the Smithsonian Institution.

Garrodia nereis, .....

4

One from the Godeffroy Museum in Hamburg.
The other three old specimens from the
Museum of the Royal College of Surgeons.

Pelagodroma marina, ....

1

Fregetta grallaria, ....

2

One a skinned trunk.

„ melanogastra, ....

2

From the Godeffroy Museum.

Procellaeiid* —

Haloeyptena microsoma,

1

(I must thank Mr. Salvin for his kindness in
allowing me to dissect his spirit-specimen
of this extremely rare bird, previously only
known from the type-specimen in the Smith-
sonian Institution.)

Procellaria pelagica, ....

3

Prosector's stores.

Cymoehorea leucorrhoa,

5

Two, one a chick, from the Smithsonian Institu-
tion. Three from Prosector's stores.

Bulweria columbina,

1

0. Salvin, Esq.

„ maegillivrayi,

1

Received from Canon Tristram (skinned trunk
only).

(Estrelata mollis, .

1

An old specimen from the College of Surgeons.

,, lessoni,

1

„ sp. inc., .

1

„ brevirostris, .

8

All young (two from the Transit Expedition).

Majaqueus ccquinoet talis,

3

One young.

Puffimis obscurus,

4

„ brem'eauda,

3

Pagodroma nivea,

1

Daption capensis,

2

One from the Zoological Society's Gardens.

Aeipetes1 antarcticus, .

2

Thalassoeca glacial aides,

1

Fuhnarus glacialis,

2

Zoological Society's Gardens.

Ossifraga gigantea,

2

One, a nestling, from the Transit Expedition.

Prion villains,

1

„ banksi,

2

One from Celebes (0. Salvin, Esq.), one from
the College of Surgeons.

„ desolatus, .

4

One a chick.

Pelecanoides urinatrix,

7

Two young.

Diomedea brachyura, .

2

„ exulans,

2

Thalassiarche culminata,

1

Pltosbetn'a fuliginosa, .

1

A nestling.

In all thirty-one species, represented by seventy-four specimens and belonging to twenty-
two different genera. Besides the above, which only represent entire birds, there were a
number of separate heads, which have been cleaned, and will be found enumerated below
in the list of osteological material in this group examined by me.

For the characters of this new genus, vide infra, p. 59.

THE VOYAGE OF H.M.S. CHALLENGER

II. PEEVIOUS LITERATURE ON THE ANATOMY AND CLASSIFICATION OF THE

TUBINARES.

I propose under this head to briefly notice the more important papers or memoirs
that have appeared dealing with the structure and classification of these birds. Titles
of several less important ones not mentioned here may be found duly recorded in the third
instalment of Dr. Coues' Ornithological Biography,1 Procellariidse, pp. 1021-1033.

1826. One of the very earliest contributions to the anatomy of the Petrels we owe
to the voyage of circumnavigation made by the " Coquille." Garnot, in the account of
that expedition,2 gives some brief anatomical notices chiefly relating to the digestive
organs of several Tubinares. The species dissected are, unfortunately, not referred to by
scientific names, but they appear to be Phcebetria fuliginosa, Thalassceca glacialoides, a
Prion, Fregetta melanogastra, and Pelecanoides urinatrix, as well as another species I
cannot determine ("Petrel de la Mer Pacificme ").

In 1827 L'herminier3 described the general character of the sternum of the Tubin-
ares, which formed his twenty-eighth family of birds, and proposed to divide the group
up, on sternal characters, into three sections — (l) the smaller Petrels (Procellaria, Cymo-
chorea, &c.) with the posterior margin of the sternum more or less entire ; (2) the
Albatrosses, with the sternum with two large and shallow excavations posteriorly; and
(3) the Petrels proper, with four posterior sternal excavations. As regards the general
position of the group, he remarks : — " Ces oiseaux . . . par la forme de l'appareil sternal,
sont intermediates aux mouettes et aux pelicans." On plate iv. of the plates illustrating
his memoir, two figures of the sternum of a Puffinus are given.

1838-39. W. Macgillivray, in Audubon's Ornithological Biography,4 describes and
figures the alimentary canal and trachea of two species of Petrels, namely, Oceanites
oceanicus (vol. v. pp. 645-646) and of Procellaria pelagica (vol. iv. pp. 313-315).

In the second part of the same author's Manual of British Ornithology5 are given a few
notes on the visceral anatomy of the British species of the group.

In the same year J. F. Brandt, in his Beitrage zur Naturgeschichte der Vogel,"
called attention to the existence of a peculiar ossicle, connected with the lachrymal and
palatine bones, and hence called " ossiculum lacrymo-palatinum," which he had discovered
in many of the Tubinares and also in Fregata aquila.

1840. It is to Nitzsch, perhaps the most acute and original ornithologist that ever lived,

1 Bull. U. S. Geol. Surv., vol. v., No. 4, Washington, 1880.

2 Voyage autour de la Monde, Zool., torn. i. ; Recherches anatomiques relatives a divers oiseaux marins,
pp. 603-612.

3 Recherches sur l'appareil sternal des Oiseaux, pp. 79-81, vol. iv., Paris, 1827.

4 Edinburgh, 1839.

5 London, 1842, pp. 258-264.

6 Beitrage zur Kenntniss der Naturgeschichte der Vdgel, St. Petersburg, 1839, pp. 4-9.

REPORT ON THE ANATOMY OF THE PETRELS. 5

that we are indebted for nearly cmr whole existing knowledge of the important subject of
the pterylosis of birds. In his classical, though posthumous, Pterylographie the Nasutse
seu Tubinares form the second group of his order Natatores, and the pterylosis of the group
is described at some length. Further details of Nitzsch's observations are recorded below
(infra, p. 14) in the space devoted to the consideration of the pterylographical characters
of these birds. So far as I am aware nothing else was contributed by Nitzsch to our
knowledge of this group.

In the same year as that in which the Pterylographie appeared, Eudolph Wagner, a
disciple of Nitzsch's, contributed to the tenth volume of Naumann's Vogel Deutschlands 1
some remarks on the anatomical structure of three genera of Tubinares included in that
work, namely, Procellaria, Fulmarus, and Puffinus. These consisted of short notes on the
skeleton and the thoracic and abdominal viscera, and, as far as they go, are accurate
enough. The general similarity in structure of the members of this group examined, as
well as of Diomedea, is noted, as well as many points of resemblance to the Laridse, and
particularly Lestris.

In the year 1844 MM. Hombron and Jacquinot communicated to the Academy of
Sciences in Paris a paper entitled " Eemarques sur quelques points de l'anatomie et de la
physiologie des Procellaridees, et essai d'une nouvelle classification de ces oiseaux."
An abstract, by the authors, is pubHshed in the Comptes Eendus for that year.2 The
material for their paper was obtained, I may remark, during the expedition of the French
ships "Astrolabe" and " Zelee," commonly known as the Voyage au Pole Sud. Basing
their classification on the form and structure of the beak, palate, and tongue, they
divide up the group as follows : —

1. Borders of mandibles excavated by a longitudinal furrow dividing them into
inner and outer cutting surfaces. Tongue small, one-third the length of beak, sagittate,
posteriorly and laterally denticulate.

Three genera — Diomedea ; Puffinus, subdivided into Puffinus proper (anglorum,
obscurus,fidiginosus'\, &c.) andPrio/inus (cinereus, aquinoctialis, arcticusV); and Thalassi-
droma (pelagica, leachii \_^leucorrhoa~\, oceanica, fregetta [= grallaria], marina.

2. Edges of upper mandible with transverse lamellae. Tongue as long as the beak,
large and thick, only free at the apex.

One genus, Prion, divided into five sub -genera — Prion s.s., Daption, Fulmarus, Ossi-
fraga, and Priocella (for Priocella gamoti= Thalassceca glacialoides of this paper).

3. Mandibles simple, with no double cutting-edges or transverse lamelke, but with two
slight, elongated " teeth " ; palate smooth or nearly so ; tongue of intermediate length.

One genus, Procellaria, separated into two groups, one with the beak quite short
(nivea, desolata, brevirostris), the other with it long (antarctica, lessoni, hcesitata, Forst.
[1 = Adamastor cinereus] ).

1 hoc. cit., pp. 555-656, 587-588, 614-617. - Loc. cit, torn, xviii. pp. 353-358.

6 THE VOYAGE OF H.M.S. CHALLENGER.

These points are illustrated in the atlas to the Voyage au Pole Sud (pi. xxxii.), the
classification adopted being explained in the text (vol. iii. pp. 143-152) published
some years later.

The genus Pelecanoides is expressly excluded from the Petrels by these authors, accord-
ing to whom it is closely allied to the Little Auk (Atte nigricans) of our northern seas.

1849. Gray and Mitchell, in the Genera of Birds,1 make the ProcellariidaB the fourth
family of their Anseres. They are subdivided into the Diomedeinae (of one genus) and the
Procellariinse, of which latter five genera are recognised (Prion, Pelecanoides, Procell-
aria, Thalassidroma, and Pufftnus). The most characteristic generic characters of these
are figured on plate 178.

1857. Bonaparte in his Conspectus2 gives a list of the then known genera and species
of the Procellariidae, which he divides into three sub-families, Diomedeinse (Diomedea),
Procellariinse, and Halodrominse. The Procellariinse again are divided into five smaller
groups, designated by letters as follows : — ■

A. Fulmarese —

Ossifraga, Fulmarus, Adamastor, Daption.

B. GCstrelatese —

(Estrelata, Cookilaria, Pterodroma, Tlialassceca, Pagodroma.

C. Prionese —

Prion, Halobcena.

D. Procellariese —

"*Unguibus comp>ressis" ; Bulweria, Oceanodroma, Thalassidroma, Pro-
cellaria.
"**Unguibus depressis" ; Fregetta, Pelagodroma, Oceanites.

E. Puffinea? —

Majaqueus, Thiellus, Nectris, Puffinus.

1866. — In this year Dr. Elliott Coues completed his Critical Review of the family
Procellariidge commenced in 1864.3

This is the most complete account yet published of the synonymy and distribution of
the species of this group, which is divided into 24 genera, containing 92 species
(17 of these being doubtful). Following Bonaparte, the same three sub-families are
adopted, the Procellariinse, as before, being subdivided into five groups.

The genera composing these are as follows : —

Section Procellariese —

Oceanodroma, Cymochorea, Halocyptena (gen. nov.), Procellaria,
Oceanites, Fregetta, Pelagodroma.

1 Loc. cit., iii. pp. 646-650. 2 Conspectus generuni avium, torn. ii. pp. 184-206.

3 Proc. Ac. Nat. Sci. Philad., 1864, pp. 72-91 (part 1), and pp. 116-144 (part 2) ; loc. cit., 1866, pp. 25-33,
(part 3), pp. 134-197 (parts 4 and 5).

REPORT ON THE ANATOMY OF THE PETRELS. 7

Section Puffinese —

Majaqueus, Adamastor, Thiellus, Nectris, Puffinus.

Section (Estrelateae —

(Estrelata, Pagodroma, Daption.

Section Prionese —

Halobcena, Pseudoprion, Prion.

Section Fulmarese —

Fulmarus, TJialassosca, Ossifraga.

Although the names of these five sections are the same as Bonaparte's, yet the genera
included in them are, it will be seen, different, the arrangement in many respects being
more natural.

1867. Eyton in his Osteologia Avium l describes briefly some of the more salient
features of the osteology of Ossifraga gigantea, Diomedea exidans and fuliginosa,
Puffinus major (and another undetermined species), and Thalassidroma oceanica
( = ? Oceanites). The skeletons of the Ossifraga, Diomedea exidans, and Thalassidroma
are figured, with details of some of the bones.

The same year witnessed the publication of M. Alphonse Milne-Edward's great work
on fossil birds.2 Pages 301-341 of the first volume are devoted to the consideration of
the osteology of the living Longipennes, composed of the Gulls (Laridse) and Petrels
(Procellariidse). The Petrels are considered to be, as regards their osteological characters,
allied most closely to the Gulls, with some resemblances to the Steganopodes. " Par
cpiekpies-uns de leur caracteres, les Procellarides se lient aux Totipalmes. Ainsi on ne
peut se refuser a reconnaitre une grande ressemblance entre la constitution de la charpente
osseuse des Fregates, des Phaetons, c'est-a-dire des Totipalmes grand voiliers et celle des
Pe'trels ou des Puffins. Cette analogie a d'ailleurs ete parfaitement saisie par L'herminier,
cpii cependant n'avait etudie que la conformation de l'appareil sternal" (loc. cit., pp.
302, 303).

A complete skeleton of Prion vittatus is figured on plate 1. fig. 1, with numerous details
of the bones of Puffinus cinereus — skull (pi. xlix. fig. 12), leg-bones (pi. li.), pelvis and
humerus (pi. Hi.), sternum and scapular arch (pi. liii.).

Hydrornis natator (pp. 362-365, pi. lvii. figs 18-22), from the Miocene deposits
of Langy, is perhaps allied to the Shearwaters (Puffinus), but the remains found (a
tarso-metatarsus, and a femur of doubtful ownership) do not suffice to decide the point
certainly.

Professor Huxley3 places the Petrels with the Gulls, Divers, and Auks in the

i London, 1867, pp. 221-225.

2 Recherches anatomiques et paleontologiques pour servir a l'histoire des oiseaux fossiles de la France, Paris,
1867-1868.

3 On the Classification of Birds, Proc. Zool, Soc, 1867, pp. 415-472.

8 THE VOYAGE OF H.M.S. CHALLENGER.

group Cecomorplise of his Sehizognathous series. Respecting their palate we read —
" The Procellariidse differ from the families which have just been enumerated
(Gulls, Divers, Grebes, Auks, and Penguins) in the great expansion of the maxillo-
palatines which become thick and spongy, and so closely approach the middle line that,
in the Albatrosses, only a very narrow cleft is left on each side of the vomer. The front
part of the vomer itself is much more strongly bent downwards than in the Gulls ; and
the ascending process of the palatine bone is greatly produced, and becomes anchylosed
with the vomer. Procellaria gigas [i.e., Ossifraga~\ holds a sort of intermediate place
between the Gulls and the Albatrosses, the maxillo-palatines being less swollen, and the
clefts between them and the vomer far larger than in Diomedea. In this species again the
basipterygoid processes are present, though I have not been able to observe them in
other Procellariidse " (loc. cit., pp. 430, 431). [As regards this last sentence, as will be seen
below, such basipterygoid processes are the rule and not the exception in this group.]
In illustration of these remarks, views of the palate of " Procellaria " gigantea and
Diomedea exulans are given.

Of the Cecomorphge, " the Procellariidse are aberrant forms, inclining towards the
Cormorants and Pelicans amongst the Desmognathae " (loc. cit., p. 458).

1871. G. R. Gray, in the Hand-list of Birds,1 places the Procellariidse between the
Uriidse and the Laridse in his order Anseres. They are divided into three sub-
families, corresponding to those already adopted by Bonaparte and Coues.

J. Reinhardt, in the same year, in his paper on the " Os crochu," or uncinate bone, in
the skull of birds,2 records its presence in nearly all the genera of this family that he has
examined. In a note on p. 339 he corrects Professor Huxley's statement as to the usual
absence of basipterygoid facets in the Petrels, such being only absent in the Albatrosses
and Procellarime (" Stormsvalerne "), present in all the rest.

1872. Carl T. Sundevall3 makes the Tubinares the fourth cohort of his order
Natatores. He adopts the same three sub-families as Bonaparte, Coues, and Gray.

1873. Reinhardt describes4 and figures two peculiar ossicles, of the nature of
sesamoids, developed at the elbow-joint of these birds in the tendon of origin of the
extensor metacarpi radialis longior. The existence of such an ossicle in the genus
Puffinus had already been described by Meckel,5 and Reinhardt finds two similar ones
developed in the Albatrosses, as well as in the genera (Estrelata, Puffinus, Majaqueus, and
Adamastor of the Procellariinse. In (Estrelata fidiginosa and bulweri, Diomedea

1 Loc. cit., vol. iii. p. 102.

2 Om en hidtil ukjendt Knogle i Hovedskallen hos Turakoerne (Musophagides, Sundev.) med nogle Bemaerkninger
om de lignende Knogler hos andre Fuglefaruilier ; Videnskab. Medd. Naturh. For. Kjobenhavn, 1871, pp. 326-341,
pi. vii.

3 Metliodi naturalis avium disponendarum tentamen, Stockholm, 1872, pp. 140-143.

4 Om Vingens anatomiske Bygning hos Storrnfugle-Familien (Procellaridse s. Tubinares), I.e., 1873, pp. 123-138;
also Gervais' Journal de Zoologie, vol. iii. pp. 139-144. 1874.

5 Traite general, &c, vol. iii. p. 144, Paris, 1829.

RETORT ON THE ANATOMY OF THE PETRELS. 9

ehlororhyncha, and Phccbetria fidiginosa, he states that the smaller of these ossicles is
wanting, though the larger is still developed. In a table he proposes the annexed
classification of the Procellariinae.

I. Wing-ossicles present. Twelve rectrices.

a. Puffineae —

CEstrdata, Puffimm, Majaqueus, Adamastor.
II. Wing-ossicles absent.

First primary longest.

More than twelve rectrices.

b. Fulmarese —

Ossifvaga, Fuhnarus, Dajation.
Twelve rectrices.

Margin of beak without lamellae.
Pagodroma.

Beak with lamellae.

c. Prioneas —

Hcdobcena, Prion.
Second primary longest.

d. Procellarieae —

Procdla via, Oceanites.

A. H. Garrod, in the same year, shows1 that the Petrels being "holorhinal" must
be separated from the " schizorhinal " Laridas and their allies. He further proposes2 to
divide the Petrels or Nasutae into two groups, the " Storm-Petrels," with a formula
AB.XY and no caeca, and the " Fuhnaridae," with formula AB.X and two short caeca.
Bulweria alone has a formula A.X and is therefore quite different from the Storrn-
Petrels. In both groups the great pectoral muscle is double, as in many of the
" Ciconiiformes," and there are two carotids.3 The Nasutas form the second cohort of his
" Anseriformes," consisting of them and of the Anseres, which latter include the Anatida?,
Spheniscidae, Colymbidaa, and Podicipitidae.

1876. P. Pavesi, in his Studi anatomici sopra alcuni uccelli,4 has given a few details
on the visceral anatomy of Diomcdea exidans, especially as regards the form of the
stomach and the presence of spines on the laryngeal eminence, continuous laterally
with a zone of similar papillae developed round the commencing oesophagus.

1 Proc. Zool. Soc, 1873, p. 37; Collected Papers, p. 128.

s hoc. cit., pp. 641, 642 ; 1874, p. 122 ; Collected Papers, pp. 204 and 220, 221. The passage on pp. 641, 642, describ-
ing the muscles of the Petrels, is unfortunately misprinted in the original paper. It is given in a corrected form, as
altered hy the writer, in the reprint of Professor Garrod's papers, p. 204. The two birds called in Garrod's text Procellaria
pdagica(\) and Procellaria fregata (?), the "Storm-Petrels" on which his observations were based, were probably in reality
Oceanites oceanicus and Garrodia nereis (cf. Proc. Zool. Soc, 1881, p. 736).

3 Loc. cit., 1873, p. 470 ; Collected Papers, p. 175. 4 Ann. Mus. Civ. Gen., vol. ix. pp. 66-S2.

(zool. chall. exp. —part xi. — 1882.) L 2

10 THE VOYAGE OF H.M.S. CHALLENGEE.

1879. Dr. Hans Gadow, in bis paper on the digestive organs of birds,1 describes tbe
alimentary viscera of tbe Tubinares, apparently based upon an examination of tbe four
genera Puffinus, Fulmarus, Procellaria, and Diomedea. Tbe arrangement of tbe intes-
tinal folds is " ortboccelic," the intestine being disposed in eigbt folds lying close to and
parallel with each other. In their ortboccelic character the Tubinares agree with the
Steganopodes and Erodii, differing from the " cycloccelic " Pelargi, Raptatores, and
Laridse.

1881. In a posthumous paper,2 published in the " In Memoriam " volume of his works,
the late Professor A. H. Garrod describes tbe anatomy of the Diving Petrel (Pele-
canoides urinatrix), based upon an examination of specimens collected during the
Cballenger's voyage. Pelecanoides has no ambiens muscle, in which respect it differs
from all the other true Petrels, and resembles Bulweria alone of them in its formula
A.X. The main vein of the leg, the femoral vein, is superficial to, instead of deep of, the
tendon of the femoro-caudal muscle, a peculiarity hitherto only observed in the genus
Dacclo amongst the Kingfishers. " The Procellariidse may be divided into tbe Storm-
Petrels or Thalassidrominse, and the true Petrels or OEstrelatinse, the former differing
from tbe latter in possessing tbe accessory semi-tendinosus muscle." These two groups
therefore correspond to those already distinguished by Garrod in his former paper as the
" Storm-Petrels " and the Fulmaridse.

As regards the systematic position of tbe Petrels it is said — " I may mention that since
writing my paper ' On Certain Muscles of Birds, and their value in classification,' I have
cbanged my views as to the affinities of the Procellariidse. In that communication
I place the family amongst the Anseriformes ; now it is evident to me that it is with the
Ciconiiformes that they are most intimately related. Reason for my change of opinion
will be found in what here follows." Unfortunately the paper was never completed, and
the reasons mentioned not stated in consequence.

In a paper read before the Zoological Society on June 1 8th of the same year 3 I
proposed to make the so-called Procellaria nereis of Gould, tbe Procellaria fregata of
Professor Garrod's earlier papers, the type of a genus to be called Garrod ia, it being not a
true Petrel at all, but one of the allied group without cseca and with a formula AB.XY,
the ThalassidromiuEe of Garrod, which includes besides the genera Oceanites, Fregetta,
and Pelagodroma, tbe family so formed constituting my Oceanitidse.

1882. Lastly, in the concluding part of the Atlas to the great work on Madagascar,4

1 Vereuch einer vergleichender Anatomie des Verdauungs-systernes der Vb'gel, Jen. Zeitschr. f. Naturw., Bd. xiii.
(n.f. vi.), pp. 92-171, 339-403, pis. iv.-ix., xvi.

2 85. Notes on the Anatomy of Pelecanoides (Puffinuria) urinatrix, loc. cit., pp. 521, 522.

3 Proc. Zool. Soc, 1881, pp. 735, 736.

4 Histoire physique, naturelle et politique de Madagascar, publiee par Alfred Grandidier, xv. ; Histoire naturelle
des oiseaux, x. i\\, Atlas iii., Paris, 1881. Plates 293, 294, 297, 298, 299, 300.

REPORT ON THE ANATOMY OF THE TETRELS. 11

MM. Grandidier and A. Milne-Edwards have given figures of the skeletons and separate
bones of Prion vittatus, Pitffinus chlororhynchus, and Thalassidroma oceanica.

III. COMPARATIVE ANATOMY OF THE TUBINARES.

My object in working out the present report has been, not to produce a detailed
description of the structure of any particular Petrel, but to describe the most important
deviations from the ordinary avian type met with in this group, and to compare the
members of it with each other, and with other groups of birds, in those points of their
structure in which experience has shown birds to differ from each other.

Some of the modifications here described are of great physiological and morphological
interest, whilst the numerous differences in points of detail displayed in the different
sections and genera of the Tubinares lead one to expect that the future study of
systematic ornithology will be not a little elucidated by the labours of the anatomist,
wherever he, as in the present case, has material at his command sufficient for something
like an adecmate study of a natural group on the basis of structural differences more im-
portant than those that can be discerned from the superficial inspection of an ordinary skin.

In the present section the external characters, pterylosis, and visceral anatomy are first
described ; these are succeeded by an account of the myology, to which follows a descrip-
tion of the tracheal structures, and of certain other points in the anatomy of the soft parts.
An account of the osteology concludes the whole.

1. External Characters and Pterylosis.

There are some points in the external characters of the Tubinares that may be noticed
here, because in ordinary skins they can only be made out with difficulty, owing to
changes and distortion in the process of drying.1

The order Tubinares derives its name from the character, prevalent throughout the
group, of the external nares, which are prolonged into a more or less Lengthy cylindrical
tube, lying usually on the dorsal surface of the beak, and opening by one or two apertures
(cf. figs. 1, 32, and 33, infra, pp. 12 and 59). The exact disposition and degree of
development of these tubes vary in the different members of the group.

In the Oceanitidae, and the smaller species of Procellariidse (belonging to the genera
Procellaria, Cymochorea, and Ilalocyptena), the nasal tubes rpuite coalesce, lying on the
dorsal surface of the beak for about its basal half; tin- tube so formed rises rather

1 I need not do more here than refer to the peculiar hill of the Tubinares, — the peculiarity arising from the sub-
division, into more or less distinct plates, of the corneous covering of the mandibles, — as it is sufficiently described in
systematic works on ornithology.

12 THE VOYAGE OF H.M.S CHALLENGER.

abruptly from the forehead, and is truncated anteriorly, the single aperture so formed
looking upwards and forward (vide PI. I. figs. 1-5). In the Oceanitidse (e.g., fig. 8) the
aperture viewed from in front is nearly circular, and with scarcely any appearance of a
median septum. In the Procellarian genera, on the other hand (fig. 9), the aperture is
more oval and distinctly double, owing to the median septum (formed by the coalesced
inner walls of the narial tubes) being much less deeply, in a lateral view, excavated
anteriorly, and so appearing to a greater extent superficially. The other Procellariinse
repeat this form of nostril, though the septum becomes much thicker, so that the nostrils
open in them by two perfectly distinct apertures (vide fig. 7, where the nostrils of
Bulweria are shown). They might thus be said to be " platyrrhine," in opposition to the
*' catarrhine " Oceanitidse and other genera already mentioned. It is in Bulweria and
Majaqueus perhaps, that the nasal septum is broadest and most superficial ; in Prion it
is well developed ; in the remaining genera it is less near the external opening, but always
quite evident. In the genus Pvffinvs the septum is also broad, but the narial tubes are
so obliquely truncated that they hardly rise above the lateral outline of the beak ; their
openings are ovals, with their longer axis vertical or (Puffinus obscurus) oblique inwards.
In Pelecanoldes the nasal tubes are short and swollen externally ; the septum is
distinct, but not broad ; and the apertures, which are sinuated ovals directed antero-
posteriorly, look almost vertically upwards, their lateral outline being nearly parallel with
the axis of the upper jaw.

In the Diomedeinre the nasal tubes are quite

separate from each other, lying just at the lower

margin of the " culminicorn." They are usually

described as tubes with a distinct circular complete

aperture, but on looking at this carefully, there

may be seen (vide fig. 1) in front and below this

tubular opening a deep cavity leading backwards

and continuous behind, over the edge of the

Fig. i.— Base of Beak of Diomcdca cxuians, to show apparent outer boundary of the aperture, with the

the form and position of the nostril. general cavity of the tube, an infolding of the outer

Avail of the latter forming the apparent outer wall of the tubular aperture.

The nasal tubes of the Petrels are formed, it may be observed, by the elongation of the
cartilaginous walls of the nasal capsules. The upper and lower turbinal cartilages are well
developed ; the alinasal turbinal cartilage, on the other hand, is represented only by a
sbght ingrowth from the internal nasal wall. Such, at least, is the condition of these
parts in Majaqueus, the only form I have examined as regards these structures.

The legs are always bare of feathers for some little distance above the tarsal joint,
the metatarsal scutellation extending upwards over the joint some little way, but disappear-
ing where the leg is covered by the feathers, and there replaced by simple skin.

REPORT ON THE ANATOMY OF THE PETRELS. 13

The scutellation of the tarsi presents different characters in the Procellariidaa and
Oceanitidae respectively. In the former, in all the forms, the legs, which are often much
compressed below the lower limit of feathering, are covered pretty uniformly hy small
scutellaa of hexagonal shape {vide PI. I. fig. 5, a). In the Oceanitidae, on the other hand,
though the back and more or less of the lateral aspects of the leg are so covered, the
front of the leg is either, as in the genera Oceanites (PL I. fig. 1, a) and Fregetta
(PI. I. fig. 4, a), " ocreate," being covered for nearly all its length by a single long scute,
or, as in Garrodia and Pelagodroma (figs. 2, a; 3, a), has a series of strong, well-
marked, obliquely transverse scutellse, extending on to the external and internal faces
of the leg for some distance.

The hallux in the Tubinares is always extremely small, and in the genus Pelccano'ides
quite absent. When present it consists only of a single joint (vide infra, p. 53, and
PI. VI. fig. 14), which, even when best developed, is very small and covered by a short,
nearly straight, spur-like claw, which projects externally, some little way above the
level of the other digits, and, being very small, may easily be passed over. In the
Oceanitidae this nail is extremely minute, considerably more so than in the Procellariidse
of similar size, but is always present1 and very straight and spur-like. In most of the
Procellariidse it is larger and more curved : it is best developed proportionately, perhaps,
in Pagodroma.

In the Albatrosses the hind-toe is so minute that these birds are usually described as
1 icing three-toed, but this is not really quite cor- led
rect. In Phcebetria the hallux externally only c^ Q &

just appears, being represented merely by a
.slight pimple-like elevation, with a very minute

claw. On dissecting away the skin, the pimple T

is seen to be connected with two minute bony
nodules, the basal one. which represents the

metatarsal, more globular, the apical One more FlG- 2. -Rudimentary Hallux of the Albatrosses, of the

natural size, except a.

pointed and covered by the minute claw. a, Phwld/.ia fttUgi. outside the skin
They are only connected by connective and nosa> sh°wing the (represented in sec-

. oi two ossicles, con- tion).

fibrous tissue to the tarso-metatarsus,- and are nected together by 6> Dicmedea exuians.

separated from each other by a considerable J^1™8 ^\ JJ" c. Diomedea h-achywa.

interspace, the whole having a total extent of covered by a minute ,/. Thaltusiarche cul-

only 3 mm. (vide fig. 2, a). ,law- wllich aw,ears ""'""'"•

In Thalassiarche (culminata) and Diomedea (brach yura and exulans) this hallux is
.still more rudimentary, and there is not a trace of a nail outside. Still, on careful

1 Mr. Dresser erroneously describes it as wanting in Oceanites (Birds of Europe, vol. viii. p. 503).

2 The existence of the rudimentary hallux in Phtebetria fuliginosa was first, I believe, pointed out by Dr. Kidder
in his account of the birds of Kerguelen's Land, Bull. U. S. Xat. Mus., vol. i. p. 22.

14 THE VOYAGE OF H.M.S. CHALLENGER.

inspection, there is a slight elevation visible on the area usually occupied by the hallux,
and on dissecting away the skin a single minute ossicle, of triangular shape, is to be
found attached to the tarso-metatarsus and surrounding structures by fibrous tissue.
This single bone probably represents the metatarsal element. In Diomedea exulans it has
a length of 0*2 inch (5 mm.) ; in the other two species, particularly in Thalassiarche, it
is much smaller, not exceeding here 2 mm. in length (vide figs. b-d). I have been unable
to find it at all in Pelecano'ides, even in quite young birds.

The anterior three toes are well developed, and are completely webbed, the web how-
ever not extending to the hallux. The claws are well developed ; in the Procellariidaj
they are always more or less curved, compressed, and sharp-pointed (vide PI. I. fig. 5, b,
Procellaria pelagica), whilst in the Oceanitidse they become lamellar, depressed and
flattened, a shape that attains its maximum in the genus Fregetta (vide figs. 1-4, b).
Hence an inspection of the legs alone of a Petrel will show, by the character of the
tarsal scutellation and the form of the claws, whether it is one of the Oceanitida3 or of
the Procellariidse.

Pterylosis. — The pterylosis of this group seems to be, on the whole, very uniform
throughout, both in the form of the tracts and the structure of the feathers.

The number of rectrices and remiges is not absolutely constant however, and there are
also some slight differences in the form of the dorsal and lumbar tracts.

As regards the number of rectrices in the Tubinares, twelve is the ordinary number
in both families, and this is never reduced,1 and only in a few instances exceeded.
Fulmarus and Daption have fourteen tail feathers, as already pointed out by Nitzsch,
and the same is the case in Thcdassceca ; Ossifraga has as many as sixteen. Aeipetes
antarcticus (in both the specimens examined by me) has, on the other hand, unlike
Thalassceca, the normal number of twelve.

The number of primary remiges is always ten, but that of the secondaries varies.
The number of these in the Oceanitidas is always ten : in the Procellariidae it is never, even
in the smallest forms (Pelecano'ides, Cymochorea, Halocyptena, Procellaria), less than
thirteen. Bidweria has twenty, which is about the average number throughout the
group, increasing however in the larger forms to twenty-nine (Ossifraga, Thalassiarche),
thirty (Diomedea brachyura), and even, in the largest of all, Diomedea exulans, to
thirty-seven.2 The pollex never has the claw so often present in birds on that digit.

As regards the distribution of the tracts of contour feathers, I may epiote Nitzsch 's
general description (Pterylography, Pay. Soc. Ed., pp. 143, 144): — "In this family the
tract-formation of Lestris is elevated into the type of a group, undergoing scarcely any
change in the form of the inferior tract, but showing some little modification in the dorsal

1 Nitzsch (Pterylogr. Ray Soc. E<1, p. 141) thought that the smaller species of Petrels had but ten tail feathers,
but such is not in reality the case.

2 Nitzsch even describes the species as having as many as forty secondaries. The total alar expanse of the specimen
I counted this number in was 9 feet 7| inch.

REPORT ON THE ANATOMY OF THE PETRELS. 15-

tract. We find, therefore, on the head a uniformly dense plumage, from which the two
principal tracts issue. The latter are separated from each other by the two lateral neck-
spaces, which extend high up, nearly to the head. The inferior tract is divided near the-
head, becomes of considerable breadth whilst stdl on the neck, and passes in this condition
on to the breast, the surface of which is covered by each band in a rather broad, parallel-
sided form, emitting no branch as far as the margin of the musculus pectoralis major.
Here it is divided by a space starting from the knee-covert in such a manner that a short
continuation of the tract, which is to be regarded as an outer branch, passes near the
knee into the lateral space of the trunk, runs on over the thigh, and soon afterwards
terminates. The other, inner branch, which represents the main band, then proceeds on
the belly, turns in a somewhat arcuated form outwards, dilates considerably in the
middle of the bow, and terminates near the anus. . . . The dorsal tract is at first
broad, becomes narrower towards the middle of the neck, then expands at the shoulder,
and divides at that point, or from the middle of the scapulas, into two limbs. In
most of the Tubinares these limbs pass uninterruptedly into the posterior half of the
dorsal tract ; and this circumstance forms their family character as distinguished from the
Longipennes. In the present group the posterior half of the dorsal tract encloses a
longitudinal space as far as the caudal pit, dilates a little outwardly on the pelvis, and
thus usually becomes united with the very oblique lumbar tracts, and grows rather strong
in the simple uropygial band, also covering the base of the oil-gland."

Nitzsch had no opportunity of examining the pterylosis of Pelecanoldes, nor any of
the Oceanitidse. His remarks were based on examination of Fulmarus glacialis, Daption
capensis, Ossifraga gigantea, Procellaria pelagica, Hcdobcena ccerulea, Puffinus obscurus,
and Diomedea exulans and cMororhyncha. Nitzsch points out certain peculiarities in the
latter genus, the most important of these being the division of the dorsal tract into two
quite separate parts — an anterior stronger part, ending in an interscapular fork, and a
posterior, weaker, dilated part. The lumbar tracts he describes as weak and uniserial. I
find this division of the dorsal tract to hold good in Diomedea exulans and brachyura, as
well as in Thcdassiarche cidminata, though the break is not very obvious, and chiefly
marked by the difference in strength of the feathers. In a nestling of Phabetria, how-
ever, there is no such break apparent ; though the dorsal tract anteriorly is stronger, it
passes behind into the posterior part, and the same condition, as is pointed out by Nitzsch,
obtains in Ossifraga. The lumbar tracts also can hardly be strictly described as uniserial,
as they tend to coalesce, by rows of interposed contour-feathers, with the external borders
of the dorsal tract, no very obvious demarcation separating the two.

Pelecanoldes and the Oceanitidse quite conform to the general type of the group, and
indeed the only at all obvious difference in this, beyond those already mentioned, lies in
the greater or less amount of the connection between the lumbar and dorsal tracts, this
being almost nil in Cymochorea and Procellaria, and considerable in the larger forms,

16 THE VOYAGE OF H.M.S. CHALLENGER.

Majaqueus, Puffinus, &e. The knee-gap may become so deep as to completely divide
the inferior tract into two parts below (e.g., Pelagodroma, Prion, and, according to
Nitzsch, Halobcena)}

The hypopteruni is usually well-developed, with long feathers, and the humeral tracts
are very strong and broad.

The contour-feathers always have an after-shaft, though in the Diomedeinse it is
extremely small, most so in Diomedea exulans where it is reduced to a short tuft, about
half an inch long, of five or six nearly simple, straight plumes. In the smaller Albatrosses
it is larger, and in the rest of the group, including Pelecanoldes, it is of good size.

All the forms have their spaces as well as tracts covered by down-feathers, which
may become very long and close-set, especially in Pagodroma.

The oil-gland is always large, globular, with its surface covered above at the base —
which is also partly covered by the termination of the uropygial band of the dorsal tract —
by scattered semi-plumes, and with a tubular mamilla, provided with a good tuft of
down-feathers. The tuft and gland are never absent. In the Oceanitidae and smaller forms
(Cymochorea, &c.) the tuft of feathers simply encircles the apex of the gland, but in the
larger ones it sends a median prolongation across it as well, so as to divide the surface
of the mamilla into two lateral parts, separated from each other by the median row of
feathers, and each with its opening or openings. The number of these varies in the
different forms of the group, as already indicated by Nitzsch (loc. cit., p. 144). Diomedea
exulans has about half a dozen small ones in each half, arranged in a crescent. Diomedea
brachyura and Thalassiarche have numerous small apertures opening into a single large
circular common opening. The Fulmars, except Aeipetes, have several apertures in each
half, as have Daption and Pagodroma, Ossifraga having as many as five. Majaqueus
has four; (Estrelata three. Aeipetes, Pelecanoldes, Bulweria, and the smaller Procellariidas,
as well as the Oceanitidae, have apparently only two pores, one in each half of the gland.

The very young birds, I may remark, are, in all the species I have seen, covered with
a thick coating of fluffy grey down, which is pushed off as usual at the ends of the
contour-feathers when the latter appear. There are apparently no intermediate changes
of plumage, the first plumage of the young bird being similar to that of the adult," a
condition of things very unlike that in the Gulls (Laridae) with which the Tubinares have
so often been associated. Besides the long down on the tracts corresponding to the
future tracts of contour-feathers, the young birds have a shorter downy covering-
distributed pretty uniformly, as in the adults, over the intervening spaces, and between
the feathers of the tracts.

1 Nitzsch lays some stress on the angle, whether acute or obtuse, made by the lumbar tracts at their junction with
the dorsal ; but the difference in the direction of the two parts is not, as seen in entire birds, so obvious as would be
judged from Nitzsch's figures (Joe. cit., pi. x. figs. 2, 3), which were probably made up from the examination of skins
only. The lumbar tracts, where the connecting rows of feathers are best developed, seem always to run outwards and
backwards from the dorsal tracts, as shown in his figure of Puffinus obscurus.

2 Diomedea exulans may be an exception.

REPORT ON THE ANATOMY OF THE PETRELS. 17

2. Alimentary Canal and its Appendages.

The Tubinares as a group agree very closely together in the form of stomach and
intestines possessed by them, which have peculiarities not occurring in any other groups of
birds, and it is only in the variations in form and structure of the tongue, in the nature of
the armature of the mucous membrane of the mouth, and in the presence or absence of
caeca that the various forms differ in any important degree from each other.

The mucous membrane of the palate usually presents, in the Tubinares, several
series of longitudinal rows of pointed, retroverted papillae, which no doubt serve in the
capture and retention of the prey by these birds. The most ordinary arrangement of
these may be understood from Plate II. fig. 21, where the palate of (Estrelata lessoni is
represented.

The palate is cleft for about half its length by a narrow median fissure, fringed on
each side by a row of small spines, which dilates behind into the opening of the posterior
nares, which is similarly fringed. Behind this, separated by a small interval, is the
linear median aperture of the Eustachian tubes. From the anterior extremity of the
median fissure runs forwards, to near the end of the beak, a sharp median ridge, with
four or five strong, conical spines developed on it posteriorly. This ridge is separated by
a deep groove on each side from the margins of the beak. Along a line corresponding to
that of the palatine bones, and extending for an extent equal to that of the median
fissure, there is developed on each side a second longitudinal row of retroverted and
pointed spines, much longer and stronger than the more median series. In front of the
posterior-narial aperture there runs between the two longitudinal series an oblique series
of smaller spines, whilst behind the Eustachian aperture is a second transverse series,
concave anteriorly.

In all the species of the genera (Estrelata, Buhveria, and Majaqucus examined by
me the same condition obtains, the larger species, however, as that figured, frequently
developing one or two rows of smaller spines lying parallel to the external longitudinal
row, one outside, and the other between it and the median fissure. Puffinus is similar,
but the anterior median keel is smooth, and almost without spines, and the palate to the
sides of the Eustachian aperture becomes covered with small spines.

In Cymochorea and Procellaria, as in Halocyptena, the palate is much as in
(Estrelata, but with all the spines smaller and feebler, particularly those on the palatal
ridge. In the first genus at least the prenarial ridge is nearly smooth, and between the
palatine row of spines — only developed posteriorly — and the median a stronger row is
developed, so that there are here altogether three pairs of longitudinal spines above.

In the Oceanitidse the palate is much the same, but the palatine row becomes very
weak and nearly obsolete, whilst the intermediate row is the strongest, considerably, of all.

(ZOOL. CHALL. EXP. PART XI. — 1882.) L 3

18 THE VOYAGE OF H.M.S. CHALLENGER.

The prenarial ridge may be slightly toothed or nearly smooth : it always ends, however,
at the commencement of the median fissure, in a slightly raised prominence, divided into
two lateral parts, in a way not seen amongst the Procellariidae. (Vide Yl. II. fig. 19,
giving an enlarged view of the palate of Oceanites oceanicus.)

Pagodroma resembles CEstrelata, but all the spines have become much smaller and
weaker, and this is still more the case in Daption, where they have almost entirely
disappeared save round the posterior nares. The line of the interior margins of the
premaxillaa and of the palatines is marked by a distinct raised ridge, and the edges of
the upper mandible, from the angle of the mouth as far forwards as the dertrum, are
marked by a series of slight, closely-set, raised ridges, oblique forwards and outwards.
It is by a great development of these that the peculiar fringed bill of the genus Prion,
reminding one of that of a duck, is produced. In Prion (t.c, fig. 23, Prion banksii)
the palate is almost smooth throughout, with the exception of a distinct prenarial
ridge, and some indications of the palatine series of spines posteriorly (not repre-
sented in the figure) : the median fissure and narial opening are however, as usual,
bounded by small spines. From a point corresponding to the angle of the mouth
forwards to a little behind where the dertrum forms the cutting edge of the bill,
the margins of the mouth are bounded by a well- developed fringe of closely-set
lamellae, reminding one much of the plates of a whale's baleen. These lamellae are
developed from the mucous membrane of the mouth, and are probably entirely epidermic
in origin ; in the cleaned skull there is no trace of their presence (vide PI. VI. fig. 4).
They are best developed a little way in front of their posterior termination of the fringe ;
here the lamellae are nearly vertical thin plates, set on at right angles to the axis of the
beak, but curved both forwards and outwards. Anteriorly they become more oblique
forwards, and much shorter. Outside of them the cutting edge of the beak is produced
downwards for a little way, so that a groove is formed between the beak and the
pectinated fringe.

When the lower bdl is in position, the more posterior and strongest of the lamellaa
completely occupy the slight space left between the cutting edge of the two jaws, lying
with their free ends curved outwards in a slight groove outside the lower mandible formed
by the reflection from it of the feather-covered skin. Anteriorly this groove disappears,
and the fringe simply lies against the outer surface — which is quite smooth, and not, like
that of the duck or flamingo, correspondingly grooved for the reception of the lamellae of
the fringe — of the lower jaw, which in front it does not even reach. In the larger-billed
Prion vittatus these lamellae are even more developed, whilst in the smaller-billed Prion
desolatus they are less so : Prion banksi is so completely intermediate in this respect
that I see no reason for the adoption of Dr. Coues' genus Pseudoprion.1 The only other

1 Proc. Ac. Nat. Sci. Phil., 1866, p. 164, where that writer has also described the structure of these fringes at
length.

REPORT ON THE ANATOMY OF THE PETRELS. 19

Petrel in which the beak is fringed in a way similiarto that here described is Halohcena
carulea, of which, however, as yet I have been unable to examine more than skins.

The existence of a peculiar fringe of lamellae along the margin of the mouth has often
been insisted on as an argument for placing Phcenicopterits amongst the Lamellirostres, but
the development of, at least, a very similar arrangement in Prion and Halohcena, birds of
a very different group again, ought to show that an adaptive contrivance of this kind
may be, apparently, independently developed without great difficulty and yet with many
features of common resemblance, in different birds without necessitating any immediate
genetic connection between its possessors.

In Fulmarus (PI. II. fig. 22) a rudimentary fringe of the upper mandible is present,
a little more developed than in Daption ; in other respects it conforms to the type of
(Estrelata, the spines, however, being very small and rudimentary. Ossifraga is similar
in all essential points, but the palate is longer, and, at least in young birds, more spinu-
lose. In Aeipetes and TJialassmca the pectination of the mandible can only just be
traced ; the palate is much longer and narrower in shape than in Fulmarus, the spines
smaller, and the palatine ridges better marked.

In Pelecanoides (PI. II. fig. 20) the palate is quite smooth throughout, with no ridges
or spines, except on the area round the posterior nares, which is pretty uniformly
covered with sharp elongated spines of fair size.

In the Diomedeinse the palate is comparatively smooth. There is a long prenarial
ridge, only with slight indications of spines at its most posterior part. The spines
bounding the narial and Eustachian apertures are well-marked, those on the palate
small and best developed towards the posterior end of the prominent palatine ridges.
Between the latter and the median fissure are developed, especially in Diomedea exidans,
additional spines of small size, as well as a short row outside their most posterior part.
Outside the tongue, between it and the inner margin of the jaws, the mucous membrane
of the floor of the mouth has on each side a well developed series of stronger spines.

The tongue, as may be seen by a glance at Plate II., where the chief variations of its
form are represented, is by no means constant in shape in the Tubinares.

In the Oceanitidse, (vide fig. 5, Oceanites), and the genera Cymochorea (fig. 9), Pro-
cellaria, and Halocyptena of the Procellariidse, the tongue is of triangular shape, fairly
fleshy basally, but tapering and becoming thinner anteriorly, its extremity being pointed
and more or less membranous, so as to easily be destroyed by rough usage. Its
posterior margin, or base, is somewhat concave, and fringed by a row of small retro-
verted pointed papillae. This is the form of tongue found, more or less modified, through
the entire group.

In (Estrelata (fig. 15),1 Majaqueus, Bulweria, and Puffinus obscurus the tongue

1 I have figured (fig. 16) on Plate II. a tongue of different form from any other known to me as occurring in the
group of Petrels. It has been labelled " (Estrelata brevirostris," but does not agree with the other species of that genus

20 THE VOYAGE OF H.M.S. CHALLENGER.

becomes more fleshy, and slightly grooved above, especially anteriorly. The sides more-
over are edged by a series of large triangular backwardly directed papilla?, extending
forwards for about the back half of the length of the tongue, but hi Puffinus to nearly
its apex. In Puffinus brevicauda (fig. 18) there are, in addition, four longitudinal fairly
regular rows of such papillae developed on its dorsal surface, those nearest the middle
line being the biggest.

In the species of the genus Diomedea (fig. 7) the tongue is also similarly covered
above, pretty uniformly, with spines, best developed on the dorsum a little behind the
apex of the organ, but is much shorter in form, being of an elongated cordate shape. In
Phcebetria (fig. 8), on the other hand, the tongue is much more pointed and elongated,
being free for about its apical two-thirds, and with the dorsal surface glabrous, the
spines being confined to its basal margin. In a nestling of this species the tongue
has much the same shape, but is covered for the greater part of its extent above by
spines, as in Diomedea: these must therefore disappear as the bird reaches maturity.
In Thalassiarche the tongue is somewhat intermediate in shape, though most resembling
that of Phcebetria. Pagodroma (fig. 14) has a very elongated, tapering tongue, with its
base and lateral margins for about their posterior quarter spinulose. The tongue of Daption
(fig. 12) is much broader and more fleshy; the spines are small, and almost confined to
its base, with only a few very obsolete ones towards the posterior angles laterally. It is
only free for a little more than a quarter of its length. Prion (fig. 13) is similar, but
the tongue is more fleshy, and the spines are smaller and quite confined to the base : the
apex is also only free for a very small extent. In Prion vittatus the tongue becomes
extremely large and fleshy, occupying the whole of the wide space between the rami of
the mandible. Aeipetes antarcticus (fig. 11) has a tongue very like that of Pagodroma,
but of course larger and less elongate : that of Thalassceca glacialoides is very similar,
but longer a little than that of Aeipetes. In all these forms the tip is blunt or emar-
ginate, with a slight dorsal groove apically. Fulmarus has a more fleshy tongue of the
same type, with a distinctly emai'ginate end, and a more evident groove, extending for
two-fifths of its length.

In Ossifraga gigantea (fig. 10) the tongue is very elongated, — three inches long, —
and narrow proportionally. Its apex is slightly emarginate, and there is a deep groove
for about two-fifths of its length, and traceable further back to the base of the tongue.
The base has a fringe of pointed spines, which are continued, of smaller size, along the
lateral margin for some way, there being some very much smaller spines developed
inside them on the borders of the tongue for about an inch, though not reaching the
posterior angles of the organ by half that extent.

Ilessoni and mollis) examined by me nor with any of my young specimens of the so-called CEstrelata brevirostris, these
resembling rather the species just named. This tongue is remarkable for having no spines laterally, those of the base
being well developed, and for its narrow and deeply grooved form and slightly emarginate tip. In spite of its label, it
belongs, I strongly suspect, to some species of the Laridse.

REPORT ON THE ANATOMY OF THE PETRELS. 21

Iii Pelecanoides (fig. 6) the tongue is fleshy, and fairly parallel-sided, tapering
apically. It is but little free, and occupies most of the interspace between the mandi-
bular rami. Its base is notched, and provided with some largish spines, which continue
forwards for about the basal half, or more, of the lateral margins. On the dorsal surface
there is alway a peculiar lanceolate mark, apparently due to a difference in the nature of
the mucous membrane covering the tongue over this area.

The oesophagus — which in the Albatrosses, as already described by Pavesi, may be
surrounded at its commencement with a zone of spines, continuous below with the spines
covering the laryngeal eminence — is always capacious and distensile, but possesses no
crop. Inferiorly, in the thorax, it passes without any marked constriction or other
difference into an enormous proventriculus, which is a thin-walled bag, reaching down
nearly to the posterior extremity of the abdominal cavity, which it largely occupies,
lying to the left side of the stomach proper and the mass of the intestines. This great
proventricular bag is twisted back on itself apically, and then, becoming slightly narrower,
passes by a small aperture into the stomach proper or gizzard. This aperture is there-
fore to the right of, and anterior to, the great " fundus," which lies freely in the posterior
part of the abdominal cavity, covering there the terminal portion of the intestine and
cloaca. Internally, the proventricular glands are seen to cover pretty uniformly the
whole surface of the mucous membrane, with the exception of a more or less narrow
zone, which lies between this glandular part and the stomach proper, correspond-
ing pretty nearly to the narrower, ascending part of the bag as seen from outside (vide
PI. II. figs. 1 and 2). The extent of this very deep "zonary" proventriculus (pr.) is
always very considerable in the Petrels, being of course, cceteris pa/ribus, larger in the
larger than in the smaller species. In Majaqaeus its extent is 4*0 inches ; in Pelecanoides,
1/85 inches : in Fregetta grallaria, 1*2 inches.

The stomach proper (g.) is always small and more or less globular, with fairly muscular
walls and provided with the usual central tendinous sheets, so that it may fairly be called a
gizzard. Its situation is peculiar, lying always above and to the right of the proven-
tricular fundus, and with its pyloric part so flexed on itself that it looks backwards
instead of forwards as in all ordinary birds (vide PI. II. figs 1, 2), in this respect
somewhat resembling the stomach of Strathio. In Struthio, however, the pyloric aper-
ture is on the deep (dorsal) side of the stomach, nearly in the middle line, and so concealed
when the viscera are viewed from the abdominal aspect. In the Tubinares the pyloric
aperture, on the other hand, is quite superficial, lying at the inferior (posterior) end of
the gizzard in the angle formed by the two parts of the bent proventriculus.

The gizzard, which is nearly always found full of the horny beaks of Cephcdopoda, is
fined internally by an " epithelium," which is usually dark in colour, and frequently of
almost corneous texture, with a more or less corrugated or wrinkled free surface (vide
PI. II. fig. 4, where the epithelial lining of the everted gizzard of Fidmarus

22 THE VOYAGE OF H.M.S. CHALLENGEE.

glacicdis is represented !). In the Oceanitidae and Diomedeinae this epithelium is softer ;
its character in other Petrels is but an exaggeration or reproduction of that existing in
some other birds, particularly that occurring in such storks as Xenorhynchus.

The displacement of the pyloric orifice of the gizzard to the left necessitates a corre-
sponding change in the commencing duodenum, so that this at first ascends in an upward
curve towards the right before it returns to form the backwardly-directed loop, character-
istic of Aves and Mammalia, round the pancreas (PI. II. fig. 1, p.).

This peculiar upward curve of the commencing duodenum, the singularly small inverted
stomach, and enormously deep proventriculus are all peculiar, so far as I am aware, to the
group of Tubinares, though universal amongst them, and no other bird yet examined
has, so far as I know, a similar disposition of these viscera.2

The intestinal caeca are entirely absent in all the Oceanitidse, but are, with one
exception, present, though of small size, in the Procellariidae. They are always short
and globular, and closely connected to the intestine, so as to appear as mere nipple-
like projections from it. Plate II. fig. 3 represents those of Majaqueus slightly en-
larged. They are usually situated quite close to the cloaca, the large intestine in nearly
all the Tubinares being quite short ; the length of the caeca themselves rarely
exceeds *25 inch, except in the very largest species (vide table, p. 23). In five speci-
mens (one a nestling) of Cymochorea leucorrhoa that I have examined, I find only
a solitary caecum, lateral in position, developed, owing apparently to the abortion of
its fellow. As Mr. Swinhoe in his description of Cymochorea monorhis3 also records
the caecum as single, it is probable that the existence of such a single caecum is
a character of the genus Cymochorea. It is not unusual, I may observe, in a group
of birds in which the caeca are of small size, and probably of no physiological im-
portance, to find specimens or species with the normal number of caeca reduced by
one. I may give as instances Mergus albellus (cf. Hunter, Observ., vol. ii. p. 325 ; and
Garrod, Coll. Papers, p. 220) amongst the Anseres, and Plotus anhinga (Garrod, I.e.,
p. 345) amongst the Steganopodes, not to mention all the Ardeidae amongst the
Herodiones. In Halocyptena, in the only specimen yet examined, I could find no trace
of any caeca at all, so that the tendency to their disappearance already observable in

1 The figure of Cams and Otto (Tabulaa Anat. Comp. Illustr., part 4, t. vi. figs. 15, 16) of the epithelium of the gizzard
of Fidmarus glacicdis does not at all faithfully represent what I have seen in two (quite fresh) specimens of that bird,
nor have I ever in other Petrels seen epithelium of such a corneous and pavement-like nature as that figured by them.
I have, therefore, had one of my specimens carefully drawn of the natural size. In this place it will be well to recall
the still more highly developed gastric epithelium of some of the Fruit-pigeons (Phcenorhina goliath and Carpophaga
latrans) described by Verreaux and Des Murs, Viallanes and Garrod (vide antca, Report on the Birds, pp. 152-154).

2 The description of these parts in the Little Auk (Alca alle) given by Professor Owen (Anat. Vert., vol. ii. p. 163),
and originally due to Home (Lect. Comp. Anatomy, i. pp. 283, 284, 1814) does not all apply to that bird (cf. the figure
and description given by Macgillivray in Audubon's Ornithogical Biography, iv. pp. 306-309), and probably refers to
some member of the Tubinares.

3 Ibis, 1867, p. 387. I have examined the type of this species, which is now in Mr. Seebohm's collection, and find
it to be a true Cymochorea.

REPORT ON THE ANATOMY OF THE PETRELS.

23

Ci/mochorea seems here to have progressed further still. The intestines are not capacious,
but the commencing duodenum may be slightly dilated.

The following are intestinal measurements : —

Small Intestine.

Large Intestine.

Ca;ca.

Total length of
Intestine.

Oceanites oceanicus, ....

100

Garrodia nereis,

—

8-5

Fregetta grallaria,

—

130

„ melanogastra, .

—

8-0

Pelagodroma marina,

—

12-2

Procellaria pelagica, . ,

8-0

1-0

•075

Cymochorea leucorrJioa, .

9-9

•6

•1

Prion desolatus,

19-5

,, banksi,

17-0

•25

•2

Daption capensis, .

33-0

1-4

(?)

Thalassceca glacialoides,

48-25

1-4

■2

Aeipetes aidarctieus,

50-2

1-2

•3

Ossifraga gigantea,

94-0

2-0

•5

Pulmonis glacialis,

53-5

1-5

•25

Puffinus obscurus, .

17-0

•5

•2

„ anglorum,

23-0

10

•2

„ brevieauda,

24-0

1-25

•25

Majaqueus cequin octialis,

54-4

1-75

•25

CEstrelata lessoni, .

420

•25

Bulweria columbina,

—

—

•25

1

2-0

PeJeeano'ides urinatrix (a),

16-25

•2

.» .. (b),

15-0

1-5

•2

Diomedea exulans,

121-4

4-0

•8

„ brachyura,

89-0

2-0

•3

The liver is usually about equilobed, the lobes not being large, and rather triangular
in shape. In the Albatrosses, however, the right lobe becomes elongated and distinctly
bigger. The gall-bladder, developed on the right hepatic duct, is always present so far
as my observations extend. The hepatic ducts (PI. II. fig. 1, r.h.d., l.h.d.) open close
together into the ascending arm of the duodenal loop, close to the pancreatic ones, of
which there are usually two or three in Majaqueus. In the specimen of Thalassceca
glacialoides dissected the left hepatic duct divided, soon after leaving the liver, into
two branches, each of which opened separately into the duodenum, so that altogether
this received three ducts from the liver. The vitelline rudiment is not to be found in
the adult birds. The bursa fabricii, in young birds at least, is a well-developed
large sac, with thick glandular walls, and a small opening into the cloaca. The spleen is
circular, or nearly so.

3. Myology.
The myology of the Tubinares presents many features of interest, as will be seen
from the following description. The species of the group, broadly speaking, resemble

24 THE VOYAGE OF H.M.S. CHALLENGER.

each other very much in the details of their muscular structure, though in the fore-
limb the structure of the biceps and the termination of the tensor patagii tendons, and
in the hind-limb the presence or absence of the ambiens and the accessories to the
femoro-caudal and semi-tendinosus, present characters available for taxonomic purposes.

Anterior Extremity.

Pectoralis primus. — This muscle is always largely developed in the Tubinares, as
might have been expected from their great powers of flight. It is peculiar in that it
is always easily divisible into two quite separate layers superimposed on each other,
besides which it gives off thin fan-like cutaneous branches. A simdar disposition of
the pectoralis primus in two distinct layers is very characteristic of many of the
Ciconiiform birds of Garrod, occurring in all the Storks and Cathartidse, and in Phaethon,
Fregata, Plotus, Sula, and Pelecanus amongst the Steganopodes. A tendency to a
similar condition, though the two layers are only separable with difficulty, may be seen
in the Ardeidse, Falconidge, and Scopus.

The superficial layer of the pectoralis primus arises {vide PI. III. figs. 1 and 2,
p. la) from the posterior and lateral margins of the body of the sternum, from the
margin of the sternal carina, and from the inferior border and external surface of the
clavicles. In the latter position it is divisible into two layers, one arising from the
extreme margin, the other and deeper from the surface, of those bones. The common
insertion into the large humeral crest is very tendinous behind, more fleshy anteriorly,
these two parts being somewhat divided by the thick tendon of the deep layer of the
muscle (vide PI. III. fig. 1, p. la).

The deep layer of the pectoralis primus arises chiefly from the body and keel
of the sternum outside the origin of the pectoralis secundus, — from which it is
separated by a strong fascia, — from the tip of the furcula, and from the fascia over the
second pectoral, especially anteriorly, where a large air-space separates these two muscles
in the interval between the furcula and coracoid (PL III. figs. 1 and 2, p. lb). Its tendon
is thin anteriorly, strong and cylindrical posteriorly, and is inserted, as already described,
between the two parts of the tendon of the superficial layer which arches over it.

The muscle is perforated a little anteriorly to its posterior border, and in front of the
strong tendinous band dividing it, by a group of vessels and nerves destined for the
supply of the muscles and skin incumbent on it. The most posterior of its fibres do not
apparently join the main tendon of insertion, but are lost in the loose fibrous tissue
occupying the axillary region.

There is a large cutaneous branch given off by the superficial layer close to its
insertion, which runs back over the humerus, and is distributed as a fan-shaped expansion
to the outer branch of the pectoral tract. Another cutaneous branch comes off from the

REPORT ON THE ANATOMY OF THE PETRELS. 2o

anterior end of the muscle on the breast close to the symphysis furculce, and goes to
the skin of tbe lower and anterior aspect of the neck.

Pectoralis secundns. — This muscle is also well developed, but though broad is
usually short, extending for not more than one-third, one-half, or sometimes two-thirds
the length of the sternum. In Pelecanoldes, however, it is much longer, extending to
nearly the end of that bone, and in Procellaria, Garrodia, Fregetta, and Pelagodroma
its extent is nearly as large, in which cases it extends beyond the posterior margin of
the deep layer of the first pectoral.

It arises from the antero-superior part of the carina sterni, and from the body of the
bone external to that, from the greater part of the coraco-furcular membrane below the
pectoralis tertius, from the symphysis furculce, and from a greater or less extent of
the antero-inferior border of the coracoid bone. Its insertion is by the usual tendon on
the superior aspect of the humerus, behind the much smaller tendon of the third pectoral.

In the Albatrosses the pectoralis secundus is unusually short, and broken up into
four cpiite separate parts, which unite before passing the shoulder-pulley. This arrange-
ment is clearly shown in fig. 2 of Plate III. representing the muscle in Diomedea brachyura.
In the other Petrels, the muscle is much more homogeneous, and only separable by
dissection into its various component parts.

Pectoralis tertius. — This muscle (PI. III. fig. 2, p. 3) is always well developed in the
Tubinares, in the form of a broad, thin band, more or less parallel with the coracoid,
occupying the superior half of the broad space between that bone and the furcula, its
fibres arising chiefly from the strong membrane between these bones, sometimes with
additions from the anterior margin of the coracoid, or from the body of the sternum
close to the middle line.

Tensor patagii brevis and longus. — These muscles have always a common, rather thin
and flat fleshy belly, arising from the extreme upper end of the clavicle, and receiving, in
addition, special small slips from the surface of the great pectoral. From this fleshy
belly spring two tendons, of which one always forms the marginal patagial tendon, and
must therefore be considered as the tensor patagii longus. Both the tendons are con-
nected, close to their origin, by fibrous slips to the humeral crest, from which indeed
they might be said to arise, receiving then the main muscular belly. The connection of
the marginal tendon with the humerus is always provided with a small tract of strong
elastic tissue (vide PL IV. fig. 7, t.p.l'.), and another such tract of longer extent is found
on its course opposite the bend of the elbow (PI. IV. figs. 3 and 7).

In other respects the development and distribution of these tendons differs much in
different groups of genera, and their arrangement will therefore be here considered
seriatim.

It is in the Oceanitidse that the disposition of the tendon of the tensor patagii
brevis (t.p.b.) is simplest, it here, in all the four genera, passing straight downwards

(ZOOL. CUALL. EXP. PART XI. 1882.) L i

26 THE VOYAGE OF H.M.S. CHALLENGER.

as a thin band, parallel to the humerus, to be lost on the fascia covering the outer side
of the forearm. In Proeellaria, Cymochorea, Halocyptena and Pelecanoides (vide
PI. IV. fig. 6) it is nearly equally simple, but as it passes over the superficial belly
of the extensor metacarpi radialis longior (e.m.) it gives off to it a small tendinous slip,
which lies on the wristward side of the main tendon.

In the genus Prion (PI. IV. fig. 1) the condition of things is slightly more complicated.
The superficial belly of the extensor m.r.L, (e.m.) is quite tendinous throughout, with
no fleshy fibres at all ; where the tensor patagii brevis (t.p.b.) crosses it the two
tendons are firmly fused together, and there is also a well-developed wristward slip sent
off from the main tendon of the tensor patagii to meet the extensor tendon
beyond this junction. The main tensor tendon where it crosses the extensor
muscle is quite free from it in most cases, though occasionally a few fleshy fibres may
arise from its anterior margin to join the deeper belly of the extensor m.r.l., (e.m).
In a specimen of Prion banksi the wristward slip goes mainly to the deep belly of the
extensor, sending off a thin band to the more superficial one. From the point of
junction of the wristward slip with the extensor tendon, a thin fan-shaped tendinous
fascia is sometimes sent off to the patagium generally.

In CEstrelata brevirostris (PI. IV. fig. 2) the condition of things is similar, but the
patagial fan is more strongly developed, and the tendinous superficial part of the
extensor metacarpi is split, proximad of the tensor patagii, one part arising superficially
to, the other (e.m.*) deep of, the prominent supracondylar humeral process.

In the genus CEstrelata proper — as represented by CEstrelata lessoni (PI. IV. fig. 4),
CEstrelata mollis, and an undetermined species — -the arrangement differs considerably from
that observed in CEstrelata brevirostris.1 The tensor patagii brevis tendon, which
is more or less fused above with the marginal tensor patagii longus tendon (t.p.l.),
develops at its junction with the superficial tendon of origin of the extensor (e.m.) —
this being, as in CEstrelata brevirostris, double — a small, elongated ossicle (a) from
which arise not only tendinous fibres — some of which form a patagial fan, whilst others
join the marginal tendon directly — but also a number of muscular fibres which form
the belly of the superficial part of the extensor. The tensor patagii brevis continues
on in the usual manner to the ulnar fascia. No bony nodule, it is to be observed, is

1 The condition above described as obtaining in CEstrelata brevirostris was exactly the same in all the specimens,
eight in number, dissected. Unfortunately all these were young birds, though the largest must nearly have attained
its mature plumage, and was probably able to fly. In other young birds in the group that I have examined the
disposition of these elbow tendons is always exactly the same as in the adults, and even when these last develop
ossicles here, such ossicles can be found, in a cartilaginous condition, in quite young birds. I have no reason there-
fore to suppose that the differences described here as existing between CEstrelata brevirostris and the other species of that
genus are due to any difference in age.

[KS. — Since the above was written, Mr. R. Ridgway has been kind enough to examine, at my suggestion, the skins
of this species in the Smithsonian Institution, and finds, as he informs me, no difference in the development of the
ossicle between this and the other species of the genus. The question, therefore, requires further material to eluci-
date it.]

REPORT ON THE ANATOMY OF THE PETRELS. 27

found iu the tendon of the superficial part of the extensor where it arises from the
humerus.

In Majaqueus (PI. IV. fig. 7), Bulweria, and Puffinus the tensor patagii brcvis
(t.p.b.) tendon is not fused with the much broader and stronger tensor patagii longus,
but is a distinct, very slight, slip, lying between this and the humerus.

At the elbow it joins the superficial ossicle (a), developed at the junction of the
tensor patagii with the extensor tendons. In Pvjffinus (brevicauda and obscurus)
this thin tensor patagii brevis is split below into two slips, one joining the deeper of
the twin tendons of origin of the superficial extensor, whilst the other is inserted on the
supracondylar process. The ossicle is larger than the corresponding one of OEstrelata, and
of somewhat smaller form ; from it spring both tendinous fibres for the patagial tendon,
and fleshy fibres for the superficial belly of the extensor (e.m.) ; from it also, or
from the fibres of the last muscle, passes off a thin tendinous fasciculus (f.) to the
ulnar fascia. Proximad of this larger ossicle is a smaller, more circular, one (a'), which
is developed in the more superficial of the twin tendons already described a little beyond
its origin, where it plays over the supracondylar process. This second ossicle is very
small in Bulweria.

In the genera Pagodroma, Daption, Fulmarus, Thalassceca, Aeipetes, and Ossifraga,
no bony nodules are developed, but the arrangement of these tendons at the elbow
becomes very complicated. Their arrangement in Ossifraga, with which the others are
almost identical, is represented in Plate IV. fig. 5.

The tensor patagii longus (t.p.l.) tendon divides near the elbow into two parts,
one continuing as the marginal patagial tendon, provided with the usual cushion of
elastic tissue opposite the bend of the arm, the other receiving the much thinner
tensor patagii brevis (t.p.b.). The united tendon so formed becomes somewhat diffused
distally, and more or less fused with the superficial tendon of origin of the extensor
metacarpi radialis longior (e.m.), from which it is continued onwards to the ulnar fascia
by two well-defined bands. Between the most wristward of these and the marginal
tendon of the patagium there is developed a narrow vinculum.

In addition to this the main tendon of the tensor patagii which has a clear, well-
defined edge on its humeral side, where it crosses the extensor muscle, sends a small
special slip of tendon (t.p'.) to the deeper of the two bellies of that muscle.

In the Diomedeinse the arrangement (PI. IV. fig. 3) more resembles that of the
Puffineas, as here also two ossicles are developed with nearly the same relationships to
their surroundings as in that group. The tensor patagii brevis (t.p.b.) is separate
from the tensor patagii longus (t.p.l.) till near the elbow, the marginal tendon of
the latter muscle having received, a little before, the very long and thin tendinous biceps
slip (b.s.).

The relations of the ossicles are very nearly as in Majaqueus (vide the figures), but

28 THE VOYAGE OF H.M.S. CHALLENGER.

the tendinous band to the ulnar fascia — which represents the morphological termination of
the tensor patagii brevis — arises in the Albatrosses nearer the middle of the fibrous
tissue lying between the two ossicles. As in the (Estrelateae and Puffinese, the tendon
of origin of the superficial part of the extensor metacarpi (e.m.) is double, and in the
figure an arrow is introduced between them to show this double nature. The proximal
and smaller of these two ossicles is developed, as before, in the more superficial of these
twin tendons. The larger of the two ossicles is somewhat different in shape in the
Albatrosses and Petrels, being more hammer-shaped in the latter group.1

The presence of these peculiar wing-ossicles is thus confined to the Dioniedeinse, and
to the genera Majaqueus, Puffinus, Buhveria, and CEstrelata (in which last there is only
one), and, according to Reinhardt (s.c, p. 133) Adamastor, of the Procellariinas. In
the genus Fregata there is a similar small bony nodule developed at the point where, as
in the Petrels, the inner part of the tensor patagii longus tendon meets the tendon of
the superficial belly of the extensor metacarpi, and from it radiate out tendinous fibres
to the patagial margin. I have observed similar ossicles, developed at points of inter-
mittent straining, in several other birds, as Larus argentatus and glaucus, Fratercula
arctica, and Merops.

These bones must be considered to be of the nature of sesamoids, which, as is well
known, are often developed in the tendons of muscles at the points of greatest strain.
Their occurrence therefore in different groups of birds is by no means a proof of any
genetic connection between such.

Biceps. — This muscle, in all Tubinares, is remarkable for its excessive reduction, the
muscular bellies being small and short, and the tendon of insertion excessively narrow and
thin (vide PL IV. figs. 1, 4, and 6, b).

It is best developed perhaps in the Diomedeinse, where as usual it arises by two
heads, a coracoid and humeral (vide PI. III. fig. 5, c, h.), both, however, being largely
tendinous, and soon uniting. From the coracoid head is given off a very narrow slip,
chiefly tendinous with a few fleshy fibres only, which runs down in the patagium, and
joins the margin of the patagium formed by the tensor patagii longus close to the
elbow (PL III. fig. 5, and PL IV. fig. 3, b.s.). '

In the Oceanitidad the biceps muscle is very slender. It has the two usual heads
of origin, the tendons of these being often closely united together by fibrous tissue,
and ending in a small short, common belly. This apparently gives off no "biceps-slip"
at all.2

1 Of. also the figures of these ossicles given by Reinhardt (s.c, p. 128).

2 The dissection of these parts in this group of birds is attended with considerable difficulty, partly owing to the
smallness of the various parts involved, partly to the great accumulations of fat round the tissues, making the true
nature of these very difficult to determine in spirit specimens. It would be very desirable to dissect out these parts in
fresh specimens.

REPORT ON THE ANATOMY OF THE PETRELS. 29

In nearly all the other Procellariidse, including Pelecanoides, the biceps becomes
modified in a peculiarly interesting way. The coracoid head alone forms the muscle
proper, whilst the humeral head, becoming detached from the coracoid head, goes entirely
to the tensor patagii longus tendon, which it joins as a short, cylindrical tendon close to
the shoulder (PI. III. fig. i, h.). It is, therefore, functionally a " biceps-slip," though it
differs from the ordinary " biceps-slip " found in so many birds,1 in that it arises inde-
pendently from the humerus, and is not a part of the true biceps muscle, although it is
supplied by the same nerve as that which goes to the coracoid head. In Diomedea, it is
to be observed, the " biceps-slip" is derived from the coracoid head alone, whereas in the
other Procellariidse this slip represents the shorter or humeral head of the normal muscle.
Only occasionally have I seen {e.g., in specimens of Procellaria pelagica, Cymochorea
leucorrhoa, (Estrelata lessoni, and Prion banksi) a very small tendinous slip derived from
this humeral head, which may be either continued downwards with the nerves and vessels
to the elbow, where it is apparently lost in the general fascia, or joins the tendon of the
true " biceps " (Procellaria, Cymochorea).

Supposing this latter to represent a more primitive condition, now nearly or epiite lost
in most of the species, the biceps muscle must originally have been two-headed, with a
patagial slip derived from its humeral head. This slip gradually increased at the expense
of the other tendon of the humeral head, till eventually the latter disappeared altogether,
the biceps proper (i.e., that flexing the forearm) being then reduced to its coracoidal
moiety.

Expansor secundariorum. — This peculiar muscle 2 is wanting altogether in the Procel-
lariidae. It occurs, however, in the Oceanitidse, though in a form different from any
previously observed, being attached to (or derived from) thoracically the surface of the
pectoralis major muscle (vide PI. III. fig. 3).

Its small belly is attached to the few last secondary remiges (S.) at the elbow, and the
thin tendon (e.s.) runs parallel to, but behind, the humerus, to the axilla, where it is joined
by a similar but shorter tendon, which is derived from the most posterior feathers of the
humeral tract, the so-called " scapularies" (Sc). The common tendon then runs for-
wards, being superficial to the extensor and flexor muscles and the nerves and vessels of
the forearm (v.n.), to be attached to the surface of the first pectoral (p. 1) close to its
insertion into the humerus. In no other instance, so far as I know, does the expansor
secundariorum become thoracically attached to the pectoralis primus, though it may
be so to the teres, coraco-brachialis longus, or coraco-brachialis brevis muscles. Nor
have I yet met with any other bird in which the tendon of this muscle is connected
to the scapularies, which here it serves to expand as well as the secondaries.

The attachment of this muscle to the pectoralis suggests that the expansor
secundariorum may originally have been formed from a cutaneous branch of the former

1 Cf. Garrod, Coll. Papers, p. 324. 2 Ibid., pp. 323-324.

30 THE VOYAGE OF H.M.S. CHALLENGER.

similar to others of the same function derived from it, which are still broad thin expan-
sions of muscular fibres. In other birds this has either completely disappeared, or has
developed thoracically new attachments to other muscles or to bone.

As regards the other muscles of the anterior extremity, the deltoid is always
remarkable for its shortness, extending but a very small distance down the arm (vide
PI. IV. fig. 7, d.) frequently allowing the anterior belly of the latissimus dorsi (l.d.)
to appear superficially below it. Only in Phcebetria fuliginosa (a nestling specimeu)
have I found the special tendinous slip of origin from the scapula which is found in
so many birds.

The triceps has a well-marked tendinous attachment to the humerus superficial to
the insertion of the latissimus dorsi. Its muscular belly arises from the scapula by fleshy
fibres, and is comparatively short, its tendon, on the other hand, being long, and not
joining the tendon of the biceps till over the elbow.

The latissimus dorsi is in two bellies, as in birds usually ; of these the posterior is
much the largest, the anterior being comparatively small and narrow.

Posterior extremity.

The gluteus primus is nearly always very small, scarcely or not at all covering
the biceps cruris (vide PL V. fig. 1). It is larger in the Oceanitidae, especially in
Oceanites (t.c, fig. 3, gl. 1) and Garrodia, where it does cover the biceps to some little
extent anteriorly.

The gluteus quintus appears to be absent, or not differentiated off from the
posterior fibres of the preceding, in all the Tubinares, except the Diomedeinas, where it
can be distinctly defined.

The ambiens is present and usually well-developed in all the Tubinares, except the
genera Fregetta of the Oceanitidae, and Pelecanoides amongst the Procellariidse, in which
it is quite absent.

In Pelagodroma, Oceanites, and Garrodia its fleshy belly is of fair size, but the
tendon I have been unable to trace across the knee, it apparently terminating on the
cnemial process of the tibia. In the other genera this tendon crosses the knee as usual,
passing in front of the patella, when that is ossified, between the great cnemial process of
the tibia and the end of the femur, and ends as usual in the leg.

The femoro-caudal is always present in the form of an usually not broad ribbon,
inserted about half way along the femur (PL V. figs. 2-4, f.c). It does not pass through,
as it does in some of the Ciconiiform birds, a sort of pulley formed by the posterior angle
of the pelvic bones.

The accessory femoro-caudal (PL V. figs. 2, 3, a. f.c.) is always present and well-
developed, except in the genera Bulweria and Pelecanoides (t.c, fig. 4), where it is quite

REPORT ON THE ANATOMY OF THE PETRELS. 31

absent. It is fairly broad and ribband-shaped, overlapping the semi-membranosus in the
Oceanitidse at its origin, and inserted into the femur together with the femoro- caudal.
In the genera Fregetta, Puffinus, and Majaqueus it is decidedly small.

The semi-tendinosus (t.c, figs. 1-3, s.t.) muscle is always present and strong. It
arises from the iliac bone round its most posterior angle, and has no connection at all
with the caudal vertebrae. Anteriorly it slightly overlaps the biceps.

Excepting in the Oceanitidse, it has no accessory head, so that all its fibres are
inserted by a thin, broad tendon, quite distinct from that of the semi-membranosus, on
to the tibia. In the Oceanitidse, in all the species and genera, there is a strong and
broad accessory head, arising from the femur, as usual, and joining the main belly of the
muscle by an oblique tendinous raphe {vide PL V. fig. 3, a.s.t.).

The semi-membranosus (t.c, figs. 1-3, s.m.) is always very large, of broad, flat,
ribband shape, not so parallel with the semi-tendinosus nor so covered by it, as usual,
its direction being more oblique than is that of the other muscle, and thus more parallel
to that of the adductors. It arises from the posterior margin of the ilium, ischium, and
pubis, from a little above the ischial prominence to within a small distance of the end of
the pubis. Its insertion is by a thin, broad tendon, anterior to, and separate from, that
of the semi-tendinosus.

Of the other muscles in the hind-limb the biceps always passes through a tendinous
loop on its way to its insertion, as is nearly always the case with birds.

The obturator externus is never large, and is inserted near the femur head. The
obturator internus is of peculiar shape, nearly oval, but with a slight indication of
becoming triangular. The deep flexors of the toes and of the hallux blend, usually about
half way down the leg, and their tendons may become ossified. Even when a hallux is
present it receives no tendon at all from these muscles.

4. Other Anatomical Features.

There are always two carotid arteries situated in the hypapophysial canal. There
are also two jugular veins, the right of which is frequently the largest.

The main artery of the leg is always the sciatic one, accompanying as it does the
sciatic nerve.

In the genus Pelecanoides, as has already been described by Garrod (cf. antea, p. 10), the
femoral vein, instead of being, as usual in birds, deep of the femoro-caudal muscle, — from
the external border of which it then seems, in the ordinary course of dissection for the
thigh-muscles, to emerge (PL V. fig. 2,f:v.), — is superficial to it, appearing at the external
edge of the obturator externus, and then crossing the femoro-caudal superficially as
represented in PL V. fig. i,/.v.

32 THE VOYAGE OF H.M.S. CHALLENGER.

In the Procellariidse, except Pelecanoulcs, the two most anterior air-cells, which lie
between the rami of the furcula at the entrance to the thorax, are not, as is usually the case
in birds, fused together to form an interclavicular air-cell, but — at least in all the species
in which I have examined into this point — remain partially distinct, being separated for
the greater part of their length by a median septum formed by the coalescence of their
internal walls — and double in consequence — but imperfect behind in the middle line,
so that there is here a free communication between the two cells over the trachea. In
the Oceanitidse and Pelecanoides the ordinary structure prevails.1

There are always large supra-orbital glands, which occupy depressions excavated for
them in the top of the skull (vide PI. VI. fig. 3), and open by a small duct into the
nasal cavities. Similar glands occur in many birds, notably the Penguins, Colymbidse,
Auks. Gulls, and many others.2

As in all other Ciconiiform birds, there is no true penis developed.

5. Trachea and Vocal Organs.

The trachea in all Tubiuares is a straight, simple tube, never convoluted in any way,
and with the normal structure of this organ in birds. In some of the genera — Fidmarus,
T/udassceca, Aeipetes, and Ossifraga — it is divided, as will be described in detail further
on, to a greater or less extent by a median longitudinal septum, as in the Penguins alone
of other birds so far as I know. The trachea has the ordinary long lateral muscle on each
side, as well as a pair of well-developed sterno-tracheales, these arising from the costal
processes of the sternum, as in so very many birds.

The constitution of the syrinx, or lower larynx, differs very considerably in the
different genera and groups of the Tubinares as regards the number and modifications of
form of the various tracheal or bronchial rings that enter into its composition. When
as, e.g., in the Gallinas, the syrinx has no intrinsic muscles, the only guides for
determination of the exact rings forming the syrinx are the variations in form of
the rings themselves, according as to whether they are tracheal or bronchial, and the
facts elucidated by a comparative study of these parts in a series of genera. Such
a study of the syrinx in the Tubinares has made it evident to me that in this
group at least the attachment of the intrinsic syringeal muscles (of which of course
there are only a pair) to a particular bronchial semi-ring is constant, thereby affording
a landmark by which the contiguous rings on both sides can at once be assigned to their
proper position. The semi-ring that bears the muscle in the Tubinares is the fifth, the
four bronchial rings (or semi-rings) above it, as well as a less or greater number of the

1 In one of the three specimens of Oceanites examined, there appeared to be a division of the interclavicular air-cell
into two, as in the Procellariidse.

2 Cf. Nitzsch's article, " Ueber die Nasendriise der Vogel," Meckel's Archiv, 1820, pp. 234-269.

REPORT ON THE ANATOMY OF THE PETRELS.

33

last tracheal rings, forming together the framework of the lower larynx. In most cases
the last tracheal ring bears a well-marked antero-posterior pessular bar.

It is in the genus Pelecanoides (figs. 3, 4) that the typical construction of the syrinx

Fig. 3. — Syrinx of Pelecanoides urinatrix, from before.1

Fig. 4. — The same, from behind.

of the Tubinares is seen in its simplest form ; it will, therefore, be described first on the
present occasion.

The last tracheal ring is complete in front, and not modified in shape ; posteriorly it
is produced downwards into a well-developed pessulus, so forming a complete three-way
piece. The first bronchial semi-rings are united in front, where they are produced
triangularly downwards ; behind, their inturned ends do not unite either with each other
or with the pessulus, or with the second semi-rings. These last, as well as the third,
fourth, and fifth, on wdiich is inserted the muscle, are all similar in shape, and separate
from each other ; they are closely approximated in front, gradually getting shorter
posteriorly.

Garrodia, which may be considered typical of the Oceanitidae, is anteriorly (fig. 5)

Fig. 5. — Syrinx of Garrodia nereis, from before.
a. The last tracheal ring, from below, to show the pessular bar.

Fig. 6. — The same, from behind.

almost the same as Pelecanoides, but the first, second, and third bronchial rings are
complete (fig. 6). The last tracheal ring bears a complete pessulus (5, a).

1 This and the succeeding figures of the syrinx of the Tubinares have been drawn as nearly as possible of one
uniform size, irrespective of that of the originals, and are also slightly diagramatic. The bronchial rings are numbered
from 1-5 ; the tracheal are marked 0, 00, 000, &c, in the reverse direction.

(ZOOL. CHALL. EXP. PART XI. 1882.) L 5

34

THE VOYAGE OF H.M.S. CHALLENGER.

A very similar type of syrinx prevails in the other Oceanitidse, and is also that found
in the genera Procellaria (figs. 7, 8), Cymochorea (figs. 9, 10), and Halocyptena. In all
these the first few bronchial rings closely resemble in character tracheal rings, being nearly
straight, closely apposed to each other, and more or less ossified. Anteriorly they may be
united with one or more of the preceding tracheal rings, and very frequently the first two,
or three are quite complete here in the middle line. There is always a well-developed

Fig. 7. — Syrinx of Procellaria pclagica, from before.

Fig. 8. — The same, from behind.

and complete pessular bar, supported behind by the last tracheal ring. "With this bar
one or more (sometimes three or four) of the bronchial rings may fuse by their coalesced
ends posteriorly, forming a broad three-way piece ; or these rings may be complete rings
closely apposed, though apparently not anchylosed, to each other in the pessular bar.
Different specimens vary slightly in the exact number and disposition of these bronchial
rings, and sometimes are not exactly similar on the two sides.

Fig. 9. — Syrinx of Cymochorea leucorrlioa, from before.

Fig. 10. — The same from behind.

In Prion vittatus (figs. 11, 12), the first bronchial ring is either small or fused with
the second, which is anteriorly entire : this is not the case with the three succeeding
ones. The last three tracheal, and first two — or on one side three — bronchial rings
form a pessular box, continuous anteriorly with the inturned anterior ends of the third
semi-rings. The fourth pair takes no share in the formation of the box. In Prion
desolatus there is only one complete bronchial ring, which may be the first, or the first

REPORT ON THE ANATOMY OF THE PETRELS.

35

and second united ; the others are incomplete, inturned anteriorly, and not co-ossified
to form a box. There is a good pessulus borne by the last tracheal.

Fig. 11. — Syrinx of Prion vitiaius, from before.

Fig. 12. — The same, from behind.

In Pagodroma (figs. 13, 14) the four first bronchial semi-rings are ossified and firmly
united into a bony box behind ; anteriorly, however, the first and fourth bronchial rings
axe free, whilst on one side the second and third are quite fused both before and behind.

The last tracheal ring is free throughout.

Fig. 13. — Syrinx of Pagodroma nivca, from before.

Fig. 14. — The same, from behind.

Daption much resembles Pagodroma, there being a bony box, formed however by
the fusion of the first three bronchials with the last tracheal ring. In neither of these
genera is there any trace of a tracheal septum.

It is by a further development of the syrinx of Pagodroma that the peculiar one of
the Fulmars is formed.

In Thalassceca glacialoides (figs. 15, 16) the last two tracheal rings are ossified and
fused together anteriorly, a median descending process being developed which lies

36

THE VOYAGE OF H.M.S. CHALLENGER.

between the similarly ossified and fused first three bronchial rings. The fourth pair of
rings is also ossified, but free from those that precede it, at least anteriorly. Posteriorly
the first four bronchial and last four tracheal rings are firmly co-ossified into a large
pessular box, whilst ossification in the median line (both before and behind) of a number
of the cartilaginous tracheal rings above this forms the line of attachment for the median
septum which divides the tracheal tube for an extent of about 1"25 inch.

In Fulmarns glacialis the syrinx is very similar ; the four bronchial rings are
anteriorly ossified, as are many of the tracheal rings in their median area. The tracheal
septum is still more developed than in Thalassceca, extending for about the lower two-
fifths of the trachea — a distance of nearly 1\ inches.

The extraordinary syrinx of Ossifraga is a still further modification of this type
(figs. 17, 18, 19). A number of the last tracheal rings (nine on one side, ten on the other,

Fig. 15. — Syrinx of Thalassceca glacialoidcs, from before.

Fig. 16. — The same, from behind.

in the specimen — a young one — figured) become completely ossified, as are the first four
or five bronchial rings, of which only the first two are complete. The inferior part of the
trachea is divided inferiorly for a short way into two quite complete and separate tubes ;
the posterior ends of the lowest tracheal rings being so much incurved that each actually,
inferiorly, forms two complete rings, those of opposite sides being quite separate, whilst
above, by the gradual diminution of their opposed interior halves, they become, when
viewed from outside, simple rings of the normal tracheal type. On section, however, it is
seen that then ends are still incurved to form a tracheal septum like that of Fulmartcs
and T/talassceca. This completely divides the trachea into two tubes for a space of about
3^ inches, terminating above by a free semi-lunar border, concave upwards (fig. 19, c, d).
In fig. 19 three sections are given of the inferior portion of the trachea made along the
lines a, b, c in fig. 17, to show how the two tracheal tubes, separate below, gradually
unite above.

REPORT ON THE ANATOMY OF THE PETRELS. 37'

Aeipetes antarcticus (figs. 20, 21), commonly placed in the genus Thalassceca with

Fig. 17.

-Syrinx of Ossifraga gigantea,
from before.

Fig. 18.— The same,
from behind.

oo

O0

Fig. 1 9. — a, b, c, sections of syrinx and trachea
of Ossifraga along the lines a, b, c, of fig.
17, to show the double nature of the
tracheal tube below, and its complete
division by a median septum above, d,
trachea opened from the side, to show the
median septum, dividing it into two paral-
lel tubes, through the left of which a
pointer is passed, below, and terminating
above by a free margin.

Thalassceca glacialoides, completely differs in the structure of its syrinx from the last
three species described, and is more like Prion. The two first pairs of bronchial rings

CD

Fig. 20. — Syrinx of Acipctcs antarcticus, from before.

Fig. 21. — The same, from behind. The smaller figure
represents a section of the trachea, to show the
imperfect septum dividing it.

are complete anteriorly, the second being ossified, for a small extent only, in the middle

38

THE VOYAGE OF H.M.S. CHALLENGER.

line. The lowest tracheal ring is quite simple anteriorly, and posteriorly it sends forwards
a pessuliform process, anchylosed in front with the second bronchial rings. Two pairs
of the bronchial semi-rings (3, 4) are ossified at their posterior extremities, but not fused
in any way. The penultimate tracheal and preceding rings are, as in Fulmarus, &c,
ossified posteriorly in their median (narrowest) portion only to bear the tracheal septum.
This, however, is not {vide fig. 18, a) a complete septum, but is incomplete, the incurved
posterior ends of the rings not reaching the anterior wall of the trachea by some little
way. Its vertical extent is small, ceasing about "85 inch above the bronchi.

The peculiar genus Bulweria is, perhaps, as far as regards tracheal structure, nearer
the small Storm-Petrels (Procellaria and Cymochorea) than any other group, judging at
least from my examination of the syrinx of Bulweria macgillivrayi. In this specimen 1
the rings are asymmetrical, — there being only three, instead of four, bronchial rings
between the pessular ring and that which bears the muscle on the left side, apparently
owing to the suppression of the second bronchial ring, as may be seen in the figures (figs.
22, 23), — and irregularly developed, tending thus to hide the typical form. The last

Fig. 22. — Syrinx of Bulweria macgillivrayi, from before.

Fiq. 23. — The same, from behind.

three tracheal rings are, anteriorly, more or less united, there being a pessular bar
developed on the inferior margin of the last ring. Posteriorly, the ante-penultimate
of these is quite free and complete ; the next is incomplete posteriorly, whilst the last is
complete on the right side, but anchyloses on the left with the first bronchial laterally.
The third bronchial rings on each side are complete, thus encircling the bronchi. The
fourth is free and incomplete.

Puffinus (as represented by Puffinus brevicauda) presents a simple form of syrinx
(figs. 24, 25), the two first bronchial rings being complete anteriorly, the succeeding
two being only semi-rings. The two last tracheal are united with the two first
bronchial behind, sending off a pessular process, which anteriorly, as usual, is continuous
with the third semi-rings. In younger birds (fig. 26) the various rings concerned
remain more distinct, the pessular bar, it is pretty clear, being largely formed by the third

1 Some of the peculiarities here described may be due to its being a youngish bird. I have, unfortunately, as yet
been unable to examine the syrinx of Bulweria colurnbina, which might throw some additional light on the subject of
the affiidties of this genus.

REPORT ON THE ANATOMY OF THE PETRELS.

39

semi-rings anteriorly, their backward prolongation fusing behind with cartilaginous
elements developed in connection with the posterior ends of the second semi-rings, and all
ultimately fusing into the pessular box of the adult. In Puffinus obscurus there is a
complete pessular bar, formed by the third bronchial rings. There is no pessular box,
the bronchial rings being all free from each other.

Fig. 24. —Syrinx of Puffinm brcvicauda,
from before.

,«~

Fig. 25. — The same, from behind. Fig. 26. — The same, from a younger bird,

in which the bronchial and tracheal
rings have not yet coalesced.

In (Estrelata lessoni (figs. 27, 28) the syrinx becomes much more specialised and
ossified. The fifth bronchial rings are strong and curved, and to these are attached, by
fan-shaped insertions of tendon, the vocal muscles. The five preceding rings (which
must therefore be the last tracheal and first four bronchial) are narrow, closely united,

Fiq. 27. — Syrinx of CEstrdata lessoni, from before.

Fig. 28. — The same, from behind.

and ossified over a rhomboidal space in the middle Une anteriorly. Behind there is a
pessular box formed by these rings, and the four preceding tracheal ones in addition,
the first two of these having a median patch of ossification. There is a well-
developed and ossified pessular bar.

40

THE VOYAGE OF H.M.S. CHALLENGER.

(Estrelata mollis is quite similar, except that the box is composed of one ring less.1
In Diomedea brachyura and exulans the calibre of the trachea diminishes very
considerably below the place of insertion of the sterno-trac/ica/es. The syrinx (figs.
29, 30) is strong and well ossified. The fifth bronchial semi-rings on which end the
fanned-out tendinous insertions of the intrinsic muscles are strong and much arched.
The four preceding bronchial rings, as well as the last tracheal, are ossified anteriorly
and posteriorly, and (with the exception in Diomedea exulans of the fourth bronchial)
co-ossified anteriorly into a strong box, with which the penultimate tracheal ring is
also connected in the middle line. Behind, the last two tracheal and first bronchial
rings are co-ossified forming a broadly triangular pessular bar or box, whilst one
or more (5) of the preceding tracheal rings have patches of ossification mesially.

Fig. 29. — Syrinx of Diomedea exulans, from before.
About natural size.

Fig. 30. — The same, from behind.
About natural size.

In Tlialassiarche cidminata (of which I have only examined one) all the five bronchial
rings, including the muscular one, are firmly co-ossified together, and free from the
rest. There is no complete pessulus, the bony box formed by tne rings terminating
posteriorly in a straight and free margin, which a cuneate bar, formed by the inturned
anterior ends of the fourth bronchial rings, does not reach. The syrinx of Phcebe1n<i
as yet I only know from a young specimen, in which the bronchial rings below the
first are incomplete anteriorly, whilst behind there is a box formed by the last two
tracheal and first four bronchial rings.

Majaqueus is very like Diomedea in its syrinx, the penultimate tracheal ring,
however, being ossified anteriorly, as well as its five successors, which remain free in

1 I regret not having as yet been able to examine any adult bird of (Estrelata brevirostris, all my specimens being
young and consequently with the tracheal rings vmossified ami generally distinct. There is a box formed by the
fusion behind of at least three of the bronchial rings with the last, or two last tracheal ones. The second, third, and
fourth bronchial rings are united together anteriorly, the third rings joining the pessular bar.

REPORT ON THE ANATOMY OF THE PETRELS. 41

front. The second bronchial ring is the last entering into the composition of the trian-
gular pessular box behind, the third and fourth rings remaining free.

In Diomedea brachyura and some of the allied species, the bronchi seem to be,
according to the late Mr. Swinhoe's notes, long and convoluted, in a way reminding one
of that which occurs in Ciconia nigra. I have not observed such convolutions in any
Albatross or other Petrel dissected by myself. I herewith append the descriptions given
by Swinhoe of this peculiarity.

Diomedea brachyura. — " In the male of this Albatross the bronchi on leaving the
trachea bulge considerably as they run horizontally, then contract, and bend forwards
and downwards, and lastly, turning sharp round, rise upwards and bulge again before
entering the lungs" (Swinhoe, Ibis, 1863, p. 431).

" A female Diomedea brachyura had the swollen and convoluted trachea which I
thought before was peculiar to the male" (i.e., 18G7, p. 227).

Diomedea nigripes. — " In this species the trachea of the female is simple, but that of
the male is terminated by large, swollen, convoluted bronchi. In a male specimen,
procured in May, the bronchi ran down right and left, almost straight for about
lT7jj inches, then took a bend forward for a short space, and narrowed, and lastly bend-
ing inwards and upwards, bulged largely and entered the lungs " (I.e., 18G3, p. 432).

In five adult males of Diomedea derogata examined, all had contorted bronchi.
These " bulge and go downwards and sidewards, then bend under the ribs on each side
into a large globe, pressing between itself and the ribs as each enters the lung at the
back" (Proc. Zool. Soc, 1873, p. 785).

6. Osteology.

M. Alphonse Milne-Edwards having described at length, with figures, the osteology of
the Tubinares in his classical work on fossil birds (vide sitjira, p. 7), whilst other points
of their osseous structure have been elucidated by Brandt, Huxley, Reinhardt, and others,
as already mentioned in the introductory part, there is not the same necessity for dwell-
ing here on this part of the organisation of the Petrels as was the case when describing the
softer and more perishable parts. Moreover, no amount of detailed description of bones,
however elaborate or well illustrated, can serve the purpose of scientific research so well
as actual specimens, which can in most cases be comparatively easily obtained for, and
permanently preserved in, museums.

My study of the osteology of the Tubinares has been chiefly based on the material
enumerated in the subjoined list.

(zool. chall. exp. — part xi. — 1882.) L G

42

THE VOYAGE OF H.M.S. CHALLENGER.

List of Osteological Material examined*

Oce.vxitiiu:.

Oceanites oceanicus, ..... Skeleton.

Fregetta melanogastra, .... Do.

Garrodia nereis, ..... Bones of trunk, and of limbs.

Pelagodroma marina, .... Limb-bones.

Proceli.aiuih.k.

Cymochorea leucorrhoa,
Proeellaria pelagica,
Bulwcria coin iiihin a,
CEstrelata " grisea" .

„ sp. inc. (also named grist us)

„ lessoni
Majagueus cequinoctialis,
Puffinus anglorum,
„ dbseurus,
„ brevicauda, .
Adamastor dnereus, .
Daption eapensis,
Aeipetes antarcticus, .
T/ialasscosa glacialoides,
Ftdrnarus glacial/*,
Ossifraga gigantea,
Prin, i desolatus,

„ vittatus,
Pelecanoides urinatrix,
Diomedea exulans,

„ brachyura,
Tlialassiarche melanqphrys,

„ cnlmiuata,

Pho'liitria fnl 'iiji no*tt ,

1

(Ey ton Coll.).

(incomplete).

Skeleton.

Do.
Limb-bones.

Skeleton.

Do.
Two skulls.
Skeleton.

Do.

Do.

Do.
Skeleton.
Skeleton (Eyton Coll.) and skull

Do. and two skulls.

Do.

Do.

Do.

Do.
Skull.

Skeleton, and sternum.
Skeleton and skull.

Do.

Do. (Eyton Coll.).
Skull.
Two skulls.

and skull.

and three skulls.

TJie skull (vide PI. VI. figs. 1-6). — The rostrum is long, and strongly-hooked apicaHy.
It has no distinct hinge with the cranium proper, but the cranio-facial notch is large,
and the nasal and premaxillary bones at their junction with, the frontals so thin as to
permit of a considerable amount of vertical movement of the beak.

The nares arc large and " holorhinal," their posterior boundaries being concave, and
not extending, by some way, as far back as the posterior ends of the nasal processes of
the premaxillse. In the Diomedeinpe the nostrils are widely separated, and distinctly
lateral in position ; in the other forms they are closely approximated, and near the
culmen. The nasal septum is but little ossified, most so in the Diomedeina?. There
may be a couple of small ossifications in the floor of the nasal capsules near their anterior
extremity, continuous anteriorly with the prernaxillse, united together in the median line,
and externally abutting on the outer and lower wall of the nostril.

REPORT ON THE ANATOMY OF THE PETRELS. 43

The skull above the orbits is always deeply excavated for the fossae in which lie the
nasal ("supra-orbital") glands. In the Albatrosses there is a strong raised external
border to the fossa posteriorly, prolonged from the post-orbital processes, whilst
externally this floor, here perforated by numerous apertures, is deeply excavated.

The post-orbital processes are large and strong. The temporal fossae are well-
developed, nearly meeting across the middle line in most, though in the Albatrosses
separated by a considerable interval. The occipital plane is inclined downwards and
forwards, but in the Albatrosses is nearly vertical. In these birds the digastric fossa; are
continuous, meeting each other in the middle line, whilst in the other Tubinares they
are separated to a greater or less extent by the wide, smooth, convex cerebellar
eminence.

As might be expected, all these fossae and their bounding ridges are much better
developed in the large Albatrosses and Petrels (Ossifraga, Majaqueus, &c.) than in the
small Procellariese and Oceanitida?, in which the skull is comparatively smooth, of much
thinner texture, and with the cerebellar eminence occupying a much greater extent
comparatively. The interorbital septum is well ossified in the larger species, most so in
Diomedea exulans, whilst in the smaller ones it is very extensively fenestrated.

The lachrymal bone is always well developed, but varies in form in the different
groups. In the Oceanitidse and the small Petrels of the ProceZfom'a-group it is T-
shaped in form, the long arm of the T being horizontal, extending forwards from
the body of the bone (which is nearly vertical) to articulate with the external
descending process of the nasal bone, just behind the level of the end of the nostril.
Between it and the rest of the skull lies a considerable oval fenestra, occupied by membrane
in the recent state. The ascending process articulates with the frontal, forming a well-
marked, backwardly-directed, antorbital process, whilst the descending process descends
downwards towards the jugal arch, to which it maybe united by ligament articulating
internally with the considerable antorbital plate of the ethmoid (" ectethmoid," Parker).

In Pelecanoides, Puffinus, Adamastbr, and Majaqueus it has the same relations, but is
more triangular in form, and closely abuts on the cranium superiorly, the fenestra being
reduced thus to a chink. In the Albatrosses it also remains separate from the skull,
and the anterior limb is but little developed as compared with the vertical part, which is
swollen, excavated by air-cells, and forms above a strong antorbital process. It loses its
connection with the ethmoid. In the remaining genera the lachrymal does not exist as
a free bone, being firmly anchylosed to the frontal above and the ethmoid anteriorly
(PI. VI. figs. 1, 3). It is hollow, with one large, and several small, external apertures.

In connection with the descending limb of the lachrymal bone.'thcre is often developed
a peculiar ossicle, named by Brandt (cf. sui^m, p. 5), who was the first to describe
its existence in Diomedea brachyura and Puffinus major, the "ossiculum lacrymo-
palatinum," from its connection with those two bones.

44 THE VOYAGE OF H.M.S. CHALLENGER.

Its nature and relations in the group have subsequently been more extensively investi-
gated by Keinhardt (vide supra, p. 8), who calls it the " os crochn."

When best developed, as in the Albatrosses, the "ossiculum lacrymo-palatinum " is a
small styliform ossicle of nearly cylindrical (as in Thalassiarche cidminata, PI. VI. fig. 7)
or somewhat lamellar (1'Iuihctria fuliginosa, PI. VI. fig. 8) shape, attached above by
an articulation to the inner face, of the descending limb of the lachrymal bone, and below
connected by a ligament to the upper surface of the palatine bone. Seen from the
side, in the dried skull (vide PL VI. fig. 1) the bone is visible below the malar arch.
It lies, in the recent state, in a cavity between the nose and the roof of the mouth,
in an oblique position, pointing downwards and inward. This bone is present in all
the genera and species of Albatrosses examined by me, as well as in Thalassiarche
chlororhyncha, as mentioned by Keinhardt, In the Oceanitidas, in Procellaria and
Cymochorea, as well as in Daption and Pagodroma, its place is taken by a narrow
ligament in which there is no ossification at all. In Bulweria, Peleca?ioides, Fulmarus,
Thalassceca, and Ossifraga there is a similar ligament, with a small, more or less ossified
nodule of bone lying in it, only connected by connective tissue with the surrounding
bones. In Aeipetes, Prion, Puffinus, Majaqucus, Adamastor, and (Estreldta it is small
and delicate, articulating with the lachrymal above, and ending freely (in the cleaned
skull) below.

It is interesting to observe that a very similar bone, both as regards shape and
position, occurs in the genus Fregata as already pointed out by Eeinhardt, whose
observation I have been able to verify. But it also occurs in forms so different from
these as the Musophagidad, many Cuculida?, Chunga and Cariama, as well as in some
Laridse and Alcidaa, so that its presence is obviously of no particular taxonomic value.
Professor Parker informs me that its precise morphological significance is at present
rather uncertain.

The palate (vide PI. VI. figs. 2 and 4) is always more or less incomplete below,
the fissure dividing it being, by the less degree of inward development of the maxillary
processes of the premaxilla?, and of approximation of the inward edges of the maxillo-
palatines and palatine bones, longer and wider in the smaller than in the larger
forms.

The maxillo-palatine processes are concavo-convex lamella?, extensively fenestrated,
pointing backwards, and with their inner edges appearing but slightly internal to the
palatine bones. They remain free from each other in the middle line, and are also
unconnected by ossification with the vomer or nasal septum. Hence the Tubinares are
in this point strictly schizognathous birds. But in the Albatrosses, where the maxillo-
palatines are very large and nearly vertical in position, the sjiace between their inner
edges is very narrow, and just in front of them the decurved end of the vomer fills up
the intervening chink, especially in Phcebetria fuliginosa, where it is firmly fixed to,

REPORT ON THE ANATOMY OF THE PETRELS. 45

though apparently not anchylosed with, the maxillary plates. The transition from this
to a desmognathons type would therefore be but very slight.

The palatines in the smaller forms are of generally flat form, with their posterior
angles rounded off, closely apposed together for a very short way behind the posterior
nares, and with fairly developed descending and ascending plates, the latter being
recurved posteriorly and anchylosed to the vomer. This latter bone (PL VI. figs. 5
and 6) is always broad behind, of generally depressed form, and strongly bent down-
wards in front, its pointed extremity appearing between the maxillo-palatinc processes at
about their anterior edsje.

In the larger forms the vomer becomes enormously broad, and keeled both above and
below. The palatines meet for a much more considerable distance posteriorly, greatest in
Fulmarus, and their descending plate becomes more pronounced ; at its anterior end the
bones of opposite sides nearly meet. The ascending plate, too, becomes very large, more
or less embracing the vomer at its base, and being separated, especially in the Albatrosses,
only by a narrow chink anteriorly from the posterior end of the equally upturned
maxillo-palatine. The posterior margin of the palatines is more or less concave. The
pterygoids are nearly straight, slightly compressed, cylindrical bones, which articulate
mesially partly with the basisphenoidal rostrum, partly with the truncated posterior
ends of the palatines. Well-developed basipterygoid facets are present in all the forms
(vide PI. VI. figs. 2 and 4), except the Diomedeina?, the Oceanitkke, Procellaria and
Cymochorea} The quadrate, as in most birds, is two-headed. Its distal end has two
distinct articular cartilage-coated areas, separated by a depression. The most external
of these is oblique from behind outwards, and is somewhat saddle-shaped, being convex
from side to side, and concave antero-posteriorly. The inner facet has its axis directed
forwards and inwards, nearly parallel to that of the pterygoid bone. It is divided by
a prominent oblique trochlea into an inner, nearly flat, surface, of triangular shape,
and a more external, deeply grooved one, also of saddle shape. As might be expected,
these features are less obvious in the feeble and smaller, than in the stronger and larger,
species of the group.

The foramen magnum is more or less reniform, with the major axis transverse, in the
small species, whereas in the biggest it is oval, especially in Ossifraga, with the long axis
vertical. The moderately sized species are here again intermediate in structure.

The mandible has no recurved angular process : its posterior end is more or less
inturned and truncated behind, the truncated surface being of triangular shape. The
articular surfaces are two in number, and, of course, of inverse shape to the corresponding
facets on the quadrate bone. One or more pneumatic foramina enter the bone at this
point.

Axial skeleton. — The number of vertebra? varies from thirty-eight to forty-two, but

1 Halocyptena has not yet been examined in this respect: it probably resembles the last two genera named.

46 THE VOYAGE OF H.M.S. CHALLENGER.

that of the cervical ones is always fifteen, as may be seen from the table appended
(p. 47). In the Oceanitidae, it will be observed, the number of cervico-dorsal vertebra;
is twenty-one, in the Procellariidae, it is twenty-two, with two exceptions, where there
are as many as twenty-three.

The articular cup of the atlas is always incomplete superiorly, the odontoid process of
the axis filling up the gap, and so completing the joint. The fifteenth cervical vertebrae
has a wTell developed free rib, which may have an uncinate process, and one or more of
the preceding vertebrae — usually two, but sometimes as many as four (Oceanites) —
have short V-shaped ribs, which do not anchylose with the vertebrae. Sometimes
(Oceanites, Prion) the fourteenth cervical rib is longer, resembling in shape that of the
fifteenth, but with no uncinate process.

The dorsal vertebrae ' are all free, except the last, or occasionally two last, which are
anchylosed to those forming the sacrum. They usually haAre well-developed hypapophyses,
especially anteriorly. These are particularly strong and well-developed in Pelecanoides
as in other diving birds (e.g., Uria, Alca, Podiceps), extending there to quite the last
dorsal vertebra. In the Diomedeinae, on the other hand, they are quite absent, or
merely represented, on the most anterior ones, by short expanded processes like those of
the few last cervical vertebrae.

In nearly all the Tubinares, each of the dorsal vertebral centra has on its sides a
distinct oval expression, of varying depth, at the bottom of which, in the largest species,
Dpi 'ii one or more small pneumatic foramina, to admit air to the interior of the bones. In
the Albatrosses, however, these pneumatic depressions are absent, though air is admitted
to the bones — which are highly pneumatic here — by a distinct, but small, aperture in
each centrum. The transverse processes, too, are in these latter birds very much hollowed
out for air cavities.

The ribs in the Oceanitidae are peculiarly broad, and flattened out dorsally, to an
extent not seen in any Procellarian.

In Pelecanoides the ribs are very long, and oblique in position, the more posterior
ones most so, with the angles formed by their vertebral and sternal moities very acute.
Thus the whole trunk almost becomes completely surrounded by a bony box, in a way
well calculated to resist the pressure of the water when these birds dive. The same
modification may be seen well-developed in the diving Alcidae (Uria, Alca, &c).

The uncinate processes are well-developed and nearly straight. They are firmly
anchylosed to the ribs.

As may be seen from the table, the number of ribs and uncinate processes varies
slightly, and the same is true for the sacral and caudal vertebrae. The latter have well-

1 I count all those vertebra; which hear ribs, whether true or false, behind the first dorsal — defined as such by
its rib being the first to articulate with the sternum — as "dorsal." The succeeding rib-less vertebrae which are anchy-
losed together are " sacral,'' the remaining free ones " caudal."

REPORT ON THE ANATOMY OF THE PETRELS.

47

developed transverse processes, and between their centres inferiorly small chevron bones,
smallest anteriorly, larger and double posteriorly, are developed. The last of these
may, apparently, anchylose with the body of the corresponding vertebra. The pygostyle
is long and compressed. The diving Pelecanoides has, it will be noticed, a greater number
of vertebrae (9) in its tail than the other forms.

Table showing the Numbers of the Vertebr.e, Ribs, and Uncinate Processes.

,

Total

Name.

Cervical.

Dorsal.

Sacral.

Caudal.

No. of
Verte-
bra;.

Ribs.

Uncinate Processes.

Oceanites oceanicus,

15

6

9

8

38

8 (2 + 6)

5 (15C. 1-4D.)

6 (15C. 1-5D.)

Garrodia nereis, .

15

6

9

0)

(?)

7(1 + 6)

Fregetta melanogastra, .

15

6

9

8

38

7(1 + 6)

5 (1 -5D.)

i -ymochorea leucorrhoa,

15

7

10

7

39

8(1 + 7)

6 (1-6D.)

ProceUaria pelagica,

15

7

9

7

38

8(1 + 7)

6 (1-6D.)

" GEstrelata grisea " (a),

15

7

11

7

40

8(1+7)

5 (1-5D.)

. " ". 9^'

15

8

10

8

41

9 (1+8)

6 (1-6D.)

Majaqueus cequinoetialis,

15

8

11

7

41

9(1 + 8)

6 (1-6D.)

Adamastor dnereus,

15

7

12

7

41

■8(1+7)

5 (1-5D.)

Puffinus anglorum,

15

7

12

8

42

8(1+7)

5(1-5D.)

,, dbscurus,

15

7

11

8

41

8(1 + 7)

5 (1-5D.)

Daption eapensis,

15

7

11

7

40

8 (1 + 7)

6 (1-6D.)

Aeipetes antarcUcus,

15

7

11

7

40

8(1 + 7)

6(1-6D.)

Thalassmca glucialoides,

15

7

11

8

41

8(1 + 7)

5 (1-5D.)

Fulmarus glacialis,

15

7

11

8

41

8(1 + 7)

5 (1-5D.)

Ossifraga gigantea,

15

7

11

7

40

8(1 + 7)

5 (1-51).)

Prim/ tJi'suhifu-1, .

15

7

10

7

39

9(2 + 7)

6 (15C. 1-5D.)

Pelecanmdes urinatrix, .

15

7

10

9

41

8(1 + 7)

6 (1-6D.)

Diomedea exulcms,

15

7

12

6

40

8(1 + 7)

5 (1-5D.)

,, bracliyiira, .

15

7

12

7

41

8(1 + 7)

5 (1-5D.)

Thalassiarche meldnqphrys,

15

7

12

7

41

8(1 + 7)

5 (1-5D.)

Pectoral arch. — The sternum (PL VII.) is usually rather broad and short, much longer
in Pelecanoides than in any other genus, with a well- developed keel, and a short, but dis-
tinct manubrium — obsolete in Pelecanoides. The costal processes are triangular in shape,
directed outwards, or in the Oceanitidse and Pelecanoides, forwards and outwards. The
anterior margin of the keel is more or less excavated, with its lower angle produced
forwards, most so in Puffinus anglorum. In Pelecanoides (PI. VII. figs. 3, 4) this part
articulates with the clavicular symphysis, instead of being merely connected to it by
ligament, as in the other forms. The coracoidal grooves are oblique backwards, and
present two distinct articular areas for the articulation of the coracoid bone. The sides
of the sternum usually converge towards the lower end of the costal process, and then
diverge again to their posterior extremities. As may be seen from the figures of Piatt' VII.

48 tup: voyage of h.m.s. challenger.

the posterior end of the sternum varies a great deal in its outline in different members
of the group.

In most of the larger forms of Procellariidae, the visceral aspect of the sternum pre-
sents, towards its anterior extremity, more or fewer pneumatic apertures, which are best
developed in the Albatrosses, where the whole bone is much permeated with air. In
Fulmarus, and all the smaller forms of Procellaxiidae, as well as in the Oceanitidaj, the
sternum has no pneumatic apertures at all, and does not contain air.

In the genera Cymochorea (fig. 5), Proeellaria, and Ilcdoeyptena, as also in Fregetta
(fig. 7) and Garrodia, the posterior margin of the sternum is entire, with only a very
slight concavity in the outline of each side. In the Oceanitine genera Oeeanites and
Pelagodroma it is very nearly the same in shape, though each side has a small excava-
tion, the margin between the excavations being convex. In Pelecano'ides (fig. 3) also
the sternum is nearly straight posteriorly. In the Diomedeinse (e.g., Thalassiarelie
melanoj)h)-i/s, fig. 1) the posterior angles of the sternum are produced backwards and
-outwards considerably, and its posterior border is broadly excavated by a sinuous curve,
■convex externally, concave mesiaUy. In Diomedea exulahs the inner concave part is
divided into two smaller concavities on each side by a process of bone, so that the
posterior margin presents four slight notches, the inner pair being the bigger.

In the remaining genera of the group, the posterior border of the sternum is always
more or less four-notched posteriorly, the notches being generally best developed and
deepest in the genera allied to Pujjhnus and Majaqueus (figs. 9-13), whilst in the
Fulmarine forms the notches are smaller and tend to be irregular. Bulweria (fig. 16)
departs widely from any of the so-called Storm-Petrels in the form of its sternum,
and approaches closely (Estrelata, &c. The exact forms of this notching, which is
inclined to vary in different specimens, may be best understood from an inspection of the
figures. The outer notch may, as in the specimen of Adamastor figured (fig. 18), be
converted into a foramen by the partial ossification of the membrane filling it.

The coracoid bones in the smaller genera (vide PI. VII. figs. 5-8) are well-developed,
with nearly cylindrical shafts, dilated internally at their distal ends to meet the acromial
process of the scapula, whilst proximally they are broad and expanded, and produced
externally into a pointed, or slightly hooked process. In Pelecanoides (figs. 3, 4) these
bones diverge at a smaller angle from each other than in the other forms ; their shafts are
less cylindrical, and the proximal ends comparatively little dilated. In the larger forms,
on the other hand, the coracoids become very divergent, and the shaft and both
extremities, but particularly the proximal one, are much dilated, so as to assume the
extraordinary form seen in Diomedea (figs. 1 and 2) and its allies, where the greatest
transverse diameter of the bone at its base is nearly as great as its entire length. The
external outline of the bone is deeply concave, owing to the great development of its
external costal process.

REPORT ON THE ANATOMY OF THE PETRELS. 40

The scapula is a slender, slightly curved bone, presenting no special peculiarities. Its-
acromial process is prolonged inwards and forwards to nearly, or quite, meet the posterior
end of the clavicle. The angle it forms with the coracoid varies very nuu-h in different
genera, being most acute in Pelecanoides, whilst in the Oceanitidse it is hardly, if at
all, less than a right angle (vide PI. VII. fig. 8, Fregetta). Procellaria and Cymochorea
resemble the other Procellariidse, the coraco-scapular angle being in them obviously
(t.c, fig. 6) acute.

The clavicles are always well-developed, strongly convex forwards as seen from the
side, and forming a more or less widely-open U when seen from in front. Their posterior
ends are produced backwards to unite, by ligament, with the acromial process of the
scapula. The symphysis is usually somewhat dilated and thickened and closely apposed
to the anteroinferior angle of the sternum, to which it is connected by ligament.
In Pelecano'iih's the clavicular arch is more V-shaped, its limbs diverging but little, and
the symphysis develops a firm articulation with the carina stem! (vide PI. VII. figs.
3, 4) as in Phalacrocorax, Sula, Plotus, and Phaethon of the Steganopodes. In
the Oceanitidse (vide PI. VII. fig. 8) the symj)hysis furcidce has a strong, curved
hypocleidial process, directed downwards and backwards, very much more developed
than the corresponding part in any Procellarian genus, even Cymochorea or Procellaria
(t.c., fig. 6).

Anterior Extremity. — The humerus (PI. VI. figs. 9-11, where that of Majaqueus
cequinoctialis is figured) is long, with a cylindrical shaft, often much conqjressed
distally in such a way that the outer margin of the bone with its condyle is
anterior and the inner one with its condyle posterior. The head is but little elevated
above the general level of the proximal end of the bone. The pectoral ridge is promi-
nent and triangular, but short, and the deltoid impression extends only as low as
its distal termination. Behind the deltoid impression is a linear one for the latissvm us
dorsi. The bicipital surface is well developed, the lesser trochanter strong and re-
curved ; it is excavated behind and below by a deep infra- capitular fossa, bounded
above by the strong interfossal ridge, the supra-capitular fossa being a more shallow
concavity. The tubercle for the insertion of the q)€Ctoralis secundus is strong, and is
situated at the commencement of the pectoral crest, just anterior to the articular
head. Below and anterior to it is an oval depression, often large, for the pectoralis
tertius. The external condyle is prolonged obliquely upwards and inwards on the
anterior surface of the bone ; the capitellum is distinct, and separated by a slight notch
from the internal condyle. The olecranar fossa is shallow and prolonged upwards into
a smooth, slightly excavated triangular area. The impression for the brachialis anticus
is deep and oval.

Above the external condyle there is a very strong, forwardly-directed epicon-
dylar process, from which arises the more superficial of the bellies of the extensor

(ZOOL. CHALL EXI\ PART XI. — 1882.) L 7

50 THE VOYAGE OF H.M.S. CHALLENGER.

metacarpi rad talis longior muscle. This is least developed in Pelecanoldes and
Thalassceca.

The pectoral crest, on its inferior aspect, presents an elongated surface for the attach-
ment of the great pectoral, coextensive with the lower moiety of its border. This surface
•develops a roughness at each extremity, particularly below, where the main part of the
more superficial layer of the muscle is attached by its strong tendon. The double nature
of the pectoral muscle is indicated by a distinct muscular ridge dividing the general area
of insertion into two.

In the Oceanitidas the humerus is conspicuously a stouter and shorter bone, with its
shaft evidently curved, instead of being almost straight. The epieondylar process
projects much less forwards, and is continued down by an elevated ridge to the surface of
the condyle itself.

In the Albatrosses the humerus is distinctly concave forwards, with its shaft consider-
ably compressed throughout. The pectoral crest is sharply pointed, the bicipital surface
very prominent and convex, the internal trochanter less developed, and the infra-capitular
fossa very shallow, with its apex occupied by a large pneumatic foramen, and the
brachialis impression long and very shallow.

In Pelecancndes, as might have been expected from its diving habits, the humerus is
somewhat modified from the ordinary Procellarian type. The shaft of the bone is
comparatively short and much compressed, especially below, where it has sharp anterior
(external) and posterior (internal) margins. The pectoral crest is little developed. The
internal condyle descends considerably lower than the external one, and the capitellar
surface is well-developed and compressed. Behind it and the external condylar trochlea
is a strongly-marked deep pit, into which fits, like a peg, a sharp conical process
developed at the proximal end of the ulna. The epieondylar process is very short, and
the depression for the brachialis anticus very shallow.

The radius is a slender, straight and cylindrical bone, with its distal end depressed
and grooved superiorly.

The ulna is much stouter, with its posterior edge sharply keeled, with only slight
impressions for the secondary remiges. The olecranon process is short and bluntly
triangular. In Pdecanoldes the radius and ulna are considerably compressed from before
backwards. The ulna is stout, and develops at its proximal end a slightly curved
triangular process, directed upwards, which, as already described, fits into a corresponding
socket on the humerus, and so firmly unites the bones together.

The manus is very long. The second and third metacarpals are nearly parallel
and straight, the third metacarpal being much more slender than its fellows. The pollex
has but one phalanx, which is strong and long, about equalling one-half of the second
metacarpal. The two phalanges of the index are well-developed, the basal one, which
does not articulate with the third digit, being much dilated posteriorly.

REPORT ON THE ANATOMY OF THE PETRELS. 5T

In the Oeeanitidse the radius and ulna are generally stouter and stronger bones
they are in the Procellariidse ; the former is considerably expanded at its distal
extremity.

As may be seen by the table at the end of this section (p. 54) the three chief segments
of the fore-limb are, as a rule, nearly equal in length, this being especially true as regards
the arm and forearm.

Pelecanoides alone has the latter much shorter than the arm, the proportions here
being three to four. In all the others the humerus and ulna are nearly equal in length.

In most of the genera the maims (excluding the carpus) is the longest of the three
segments, but this is not the case in the larger forms (Adamastor, Majaqueus, and
Otssifraga) of the Procellariinae, whilst in the Diomedeinae the manus is very much
shorter, as may be seen by the measurements, than either the humerus or ulna.

Pelvic arch. — The pelvis (cf. PI. VI. fig. 12, pelvis, &c, of Majaqueus cequi-
nocticdis) may be described as generally elongated and narrow. The prce-acetabular is
about equal to the post-acetabular axis, though in Cymocliorea, Procellco'ia, Pelecanoides
and the Oeeanitidse it is considerably longer. In Puffinus, on the other hand, the reverse
is the case.

The ilia are long and narrow ; anterior to the acetabula they are slightly concave
plates, with their anterior extremities somewhat rounded off, separated mesially by the
sacral vertebrae, the neural spines of which coalesce into a strong median ridge The
antitrochanteric eminences are strong, and stand out conspicuously, the iliac bones
attaining here their greatest transverse extent, though each bone is narrow and
separated by a wide space, occupied by the bodies and transverse processes of the more
posterior sacral vertebras, from its fellow of the other side. A strongly marked post-
acetabular ridge runs from here inwards and backwards to the prominent posterior iliac
angle, which lies between the transverse processes of the second and third caudal verte-
brae. External to the ridge, the iliac surface is nearly vertical.

The ischia are narrow and compressed plates of bone, usually strongly curved down-
wards posteriorly to articulate with the dilated posterior ends of the nearly straight,
slender, pubic bones, each of which has at the level of the anterior angle of the acetabu-
lum, a slight prepubic spine. The posterior ilio-sciatic margin is first strongly concave
backwards, and then convex.

Seen from below, the pelvis preserves its generally narrow shape, the pubes being
only slightly inturned at their posterior, cartilaginous extremities. The renal fossae are
narrow, fairly deep and confluent. In front they are limited by the transverse pro-
cesses of the 3rd or 4th sacral vertebras, which, like those of their predecessors, are
short and slender, the 7th, 8th, 9th and 10th of those bones developing transverse
processes, which abut against the ilia, and in the larger forms become strong and more
or less double.

52 THE VOYAGE OF H.M.S. CHALLENGER.

There is not very much divergence from the general form of pelvis described here
amongst the various forms of Petrels. In the Albatrosses it becomes very narrow,
especially anteriorly, the renal fossae being correspondingly narrow and deep. The
bones entering into its composition, and supporting it become extensively pneu-
matic. Anteriorly the ilia unite, or nearly so, over the neural crest, whilst pos-
teriorly the foramina between the transverse processes of the sacral vertebras become
reduced to small holes, owing to the increased amount of ossification. A tend-
ency to similar increased ossification in these parts is discernible in the larger forms
of Petrels.

In Pelecano'ides the ischia are nearly straight along their posterior margin.

In Ci/rnochorea and Procellaria, as well as in the Oceanitidas, the pelvis generally is
weaker, with its posterior region more deflexed. But that of the two former genera may
be distinguished from that of the Oceanitidaa by the obsolete condition of the posterior
iliac spine, the weakness of the post-acetabular ridge, and the smaller size of the ilio-sciatic
and obturator foramina.

Posterior Extremity. — The femur is short, and more or less curved, most so in
Puffinus. The head is a little elevated, and set on nearly at right angles to the long
axis of the bone. The external condyle descends lower than the internal, and has a
well-marked posterior trochlear surface. The femur of the Oceanitidse is a stronger and
better developed bone, particularly at the two extremities, than it is in the Procellariidaa
of corresponding size. The tibia is long, especially in the Oceanitidaa, where it is at
least twice as long as the humerus. It has a very large and strong cnemial process,
best developed in the genus Puffinus, for the attachment of muscles, rising high above
the articular faces of the femur {vide PI. VI. fig. 13). Its distal end is more or less
antero-posteriorly compressed, and has an osseous bridge for the extensor communis
digitorum. The ridge for the fibula extends for about one-quarter the total length of
the shaft of the bone, beginning a little below the external condylar facet.

The fibula is a delicate, styliform bone, which may be two-thirds as long as the
shaft of the tibia.

The tarso-metatarsus has a smooth, rounded, interarticular prominence proximally,
and a calcaneal eminence, with two deep grooves, which may become converted into
canals, for the passage of the flexor tendons. In the Diomedeinae this calcaneal process
is feebly developed, with but a single groove internally, and a broad trochlear surface,
with two shallow furrows, externally. The antero-external margin of the bone is sharply
keeled. In such forms as Puffinus this keel becomes very sharp and prominent, owing
to the greatly compressed form of the leg. Internal to it the bone is distinctly
grooved.

Of the articular trochlear at its distal end, the inner is the shortest and most
oblique. Those for the third and fourth digits are more nearly equal, that for the

REPORT ON THE ANATOMY OF THE PETRELS. 53

third however being slightly the longer. There is a small foramen between it and
the fourth.

In Puffinus there is a distinct osseous bridge, developed on the anterior and distal
surface of this bone, external to the impression for the tibialis anticus, which I have
also seen present (on one side only) in Diomedea exulans. Usually the bridge remains
tendinous.

The three anterior digits are strong and well developed, the third and fourth being
nearly equal in length. They have the normal number of phalanges, of which the
basal one is always much the longest. In the Oceanitidas the phalanx of the middle
digit always exceeds the two succeeding ones, taken together, in length, whereas in the
Procellariidse it is always shorter, considerably, than these two.

The hallux is altogether absent in Pelecanoides, and is present only in the most
rudimentary form, as already described {supra, p. 13) in the Diomcdeina;. In the
Oceanitidas and remaining Procellariinse it is always present, though small, but is
peculiar in consisting of only a single phalanx, which bears the claw (vide PI. VI. fig. 14).
It articulates, proximally, with a small metatarsal, wdiich lies in its usual relationship to
the cannon-bone formed by the conjoined metatarsals.

In the ordinary Petrels the only pneumatic bones of the skeleton are the skull, lower
jaw (around its angle), sternum (very slightly), and the cervical, dorsal, and some of
the more anterior sacral vertebras. The limb bones are all filled with marrow. In the
smaller forms indeed of both families only the skull, lower jaw, and a few of the most
posterior cervical vertebras seem to be pneumatic. As a rule there seems to be a gradual
increase in the amout of pneumaticity of the bones correlated with the increase of size in
the bird generally.

In the Albatrosses the whole of the axial skeleton (excepting some of the ribs,
the scapula, furcula, caudal vertebras, and uncinate processes) becomes extensive!)* pneu-
matic, the sternum being especially so. The humerus, moreover, becomes hollowed and
filled by air, which enters through the pneumatic foramina developed at the bottom of the
infra-capitular fossa.

The proportion of the hind, as compared with the fore, extremity, as well as those
between different segments of those limbs, are very different, as may be seen from the
appended table of measurements (in millimetres), in the Oceanitidas and the Procellariidas
respectively. In the former the leg, as measured by the combined lengths of the femur,
tibia, and metatarsus, and therefore excluding the toes, is longer than the wing (humerus
+ ulna + manus (omitting the carpals). The tarsus is longer than the mid-toe or ulna,
and at least twice as Ions; as the femur. The tibia is at least twice as lona; as the
humerus, and much longer than the manus.

In the Procellariidas (including the Diomedeinas and Pelecanoides) the leg, measured
in the same way, is shorter than the wing. The tarsus is not longer than the mid-toe

54

THE VOYAGE OF H.M.S. CHALLENGER.

(except in Procellaria where it is just) but shorter, and the same is always the case
when it is compared to the ulna. It is never twice as long as the femur. The tibia
is only a little, or not at all, longer than the humerus or manus.

Name.

Humerus.

Ulna.

Manus.

Femur.

Tibia.

Tarso-
■Meta tarsus.

Middle
Toe.

Oceanites oaeanicus,

23

21

34

15

50

35

28

Garrodia nereis, .

20

17

26(?)'

14

51

34

25

Pelagodroma marina, .

27

24

37

18

60

41

35

Fregetta melanogast ra , .

27

23

36

19

56

38

26

Cymuchorea leucorrhoa,

35

35

42

16

37

24

26

Procellaria pdagica,

26

24

33

13

33

22

20

Bulweria columbina,

62

62

63

20

42

28

28

" (Estrelata grisea,"

81

83

84

31

61

36

43

M< ijt iqueus ceqii inoct i< I lis,

151

154

143

51

116

67

81

AdamastoT cinereus,

134

132

127

50

108

62

78

Puffinus anglorum,

79

72

86

31

79

46

51

„ obscurus,

66

63

71

25

66

40

44

Daption capensis,

86

84

91

38

80

46

56

Ao petes antarcticus,

98

93

100

44

88

44

56

Thalassceca glacialoides,

115

113

118

48

96

57

68

Fulmarus glacialis,

118

116

117

50

113

54

67

Ossifraga gigantea,

243

236

212

88

184

94

130

Prion, desolatus, .

57

56

56

23

53

32

35

Pelecanoides urinatrix,

43

33

44

23

46

24

27

Diomedea exulans,

428

417

290

110

246

124

168

,, braehyura, .

281

285

222

76

175

96

122

TJtalassiarcJte melanopkrys,

259

262

202

80

161

83

118

IV. THE CLASSIFICATION OF THE TUBINARES.

The propriety of the division of the entire order Tubinares into two main families,
which must be termed the Oceanitidae and ProceUariidse,3 first proposed by Professor
Garrod in 1873 (vide supra, p. 9), has been fully borne out by my further investigations
into the structure of these forms. To the differences in their myological formulae, and in
the presence or absence of caeca, may now be added numerous other points, both external
and internal.

The Oceanitidae agree together in having the following peculiarities which are not
shared in — with one or two exceptions marked by an* — by any of the Procellariidae : —

The number of secondary remiges is never more than ten. The tarsi are not uni-
formly reticulate, but are either ocreate, or covered by large transversely-oblique

1 Imperfect in the specimen measured. This length is estimated.

2 Cf. Proc. Zool. Soc, 1881, p. 737.

REPORT ON THE ANATOMY OF THE PETRELS. 55

scutes anteriorly. The claws are very flat, depressed, and lamellar. There are no colic
caeca.* (Absent in Ilalocyptena only of the Procellariidae.) There is a peculiar expansor
secundarioruni muscle. The tendon of the tensor patagii hrevis is quite simple
throughout. The semi-tendinosus muscle has a well-developed accessory head. The
ambiens muscle, when present, does not pass over the knee, but is lost on the cnemial
process of the tibia. The number of cervico-dorsal vertebrae is twenty-one. The
clavicles have a long, curved, symphysial process. The leg bones are longer than the
wing bones. The tarsus is longer than the mid-toe* and ulna, and at least twice as long as
the femur. The tibia is at least twice as long as the humerus, and much longer
than the manus. The basal phalanx of the middle toe is as long as, or longer than, the
next two taken together.

The Oceanitidae also agree together in having no basipterygoid processes, no uncinate
bone, a peculiarly short and stout humerus, radius, and ulna, a single circular nasal
aperture, a sternum with its posterior margin quite or nearly entire, a larger gluteus
primus, as well as in numerous other smaller details already noticed. All these
characters never coexist together in any Procellarian form, and, if my observations are
correct, the Oceanitidae further differ from the Procellariidae by having a biceps brachii
muscle of the normal form, with no patagial slip.

The Procellariidae on the other hand, have the following characters : —
The number of secondary remiges is never less than thirteen, and is usually much
greater. The tarsi are pretty uniformly covered with small hexagonal scutella?. The claws
are sharp, curved, compressed. Short colic caeca are present.1 There is no expansor
secundarioruni muscle. The termination of the tendon of the tensor patagii hrevis
is never quite simple, and may become very complicated. There is no accessory head
to the semi-tendinosus. The ambiens muscle (only absent in Peleanoldes) always
crosses the knee. The number of cervico-dorsal vertebras is not less than twenty-two.
The clavicles have only a very small symphysial process. The leg is shorter than the
wing. The tarsus is not larger than the mid-toe (except in Procellaria) , and is shorter
than the ulna. It is never twice as long as the femur. The tibia is only a little, or not
at all, longer than the humerus or manus. The basal phalanx of the middle toe is shorter
than the two next joints. Basipterygoid facets may or may not be present, and the
same is true of the uncinate bone. The humerus, radius, and ulna have a shape different
from that of the Oceanitidae. The form of the nostrils, and of the posterior margin of the
sternum, varies extensively. The gluteus primus is always very small, and there is a
peculiarly formed patagial slip derived from the biceps muscle.

1 Halocyptena is apparently an exception to this rule, but as Cymochorea has only one caecum, there is nothing
surprising in the reduction being carried a step further. As therefore all the congeners of Halocyptcna have ea-ca, it
may be safely assumed that their disappearance in it has been very recent, and has occurred since it acquired the
rest of its Procellarian characters. This loss of ca;ca therefore by it does not in any way really approximate it to the
Oceanitidie.

5G THE VOYAGE OF H.M.S. CHALLENGE!:.

Tims in spite of the genera] superficial resemblance of the Oceanitidse to the smaller
forms of Procellaxiidae, with which all oruithologists previous to Garrod had confounded
them, the differences between the two families are, it will he seen, numerous and
important. The special points of resemblance which the Oceanitidse have with such
Procellarian genera as Procellaria and Oymochorea — such as the general small size,
style of coloration, form of skull, comparative simplicity of the tensor patagii arrange-
ment, simple sternum and syrinx (the last three peculiarities being also common to
Pelecanoides) — may best be explained by supposing that these small Procellarian forms
are on the whole less specialised than the larger ones (Fulmars, Albatrosses, Shearwaters,
&c), and so retain more of the characters possessed by the primitive and now extinct com-
mon form from which both the Procellariidaj and Oceanitkhe must have been derived.

The Oceanitidas are a small and, on the wdiole, compact group, with but few differences
of importance between the four genera contained in it. Of such differences the most
important are the loss of the ambiens, and the very flattened nails and feet of
Fregetta ; the lengthening of the foot in Pelagodroma ; and the acquisition of an
ocreate tarsus by Fregetta and Oceanites. Garrodia is, therefore, on the whole, the least
modified form of the group. The four genera may be distinguished as follows : —

Garrodia. Ambiens present ; tarsus scutellated anteriorly ; sternum posteriorly
entire.

Oceanites. Ambiens present ; tarsus ocreate ; sternum posteriorly slightly excavated ;
interdigital webs yellow.1

Pelagodroma. Ambiens present; tarsus scutellated; sternum and webs as in
Oceanites; feet very long.

Fregetta. Ambiens absent; tarsus ocreate; sternum entire; feet very short, and
nads peculiarly broad and blunt.2

The Procellariidas, comprising as they do by far the greater number of species and
genera of the group, show much more divergence inter se than is the case with the
Oceanitidas. The Albatrosses are by far the most aberrant forms of this group, with
which, however, they have all the characters above noted in common, though in them-
selves specialised in several points. The discovery of a rudimentary hallux, and of an
aftershaft in these birds, disposes of two of the characters which have hitherto been
available for their separation from the other Tubinares, as do the gradations of form
that exist in the amount of separation of the two parts of the dorsal tract of another.
As peculiarities of the Diomedeinse may be included : — ■

1 To the genus Oceanites belong Thalassidroma gracilis (Elliott, litis, 1859, p. 391 — the type (?) of which, now in
the Smithsonian Institution, I have examined) and Thalassidroma lineata (Peale, Orn. U.S. Expl. Exped., pi. xxxix.
p. 403). Thalassidroma segethi (Philippi and Landbeck, Wiegm. Arch., 18u'0, p. 282) may be the former bird, or, as
suggested by Mr. Salvin (Proc. Zool. Soc, 1878, p. 736), Fregetta grallaria.

0 Besides Fregetta grallaria and melanogastra there seem to be two other species to be referred here, viz., Procellaria
albogidaris, Finsch, (Proc. Zool. Soc, 1877, p. 722), and Fregetta mcestissima, Salvin (Proc. Zool. Soc, 1879, p. 130).

REPORT ON THE ANATOMY OF THE PETRELS. 57

The Literal position of the nostrils.1 The presence of a distinct gluteus quintus
muscle. The formation of the biceps humeri muscle, which gives off a patagial slip
from its coracoidal head. The characteristic sternum. The absence of haemapophyses
on the dorsal vertebras. The pneumatic os humeri. The generally pneumatic condition
of the skeleton. The proportion of the manus to the humerus and ulna.

The tongue and palate are also more or less peculiar, and in all the genera there are
uncinate bones, no basipterygoid facets, and two large distinct accessory wing-ossicles ;
the right liver-lobe is also distinctly the larger of the two.

There are apparently three good genera of Albatrosses which may be distinguished,
independently of external characters, as follows : —

Diomedea. Tongue very short ; uncinate bones more or less styliform. (Diomedea
exulans and brachyura.)

Tludassiarche. Tongue intermediate ; uncinate bones styliform. {Thalassiarche
culminata.)

Phabetn". Tongue much longer; uncinate bones flattened; hallux better developed
than in the other genera, and with an external claw. (Phoebetria fidiginosa )

Neglecting for the present the peculiar diving Peleeano'ides, the remainder of the
Procellariidae forms a natural group distinguished by the following characters from the
Albatrosses (Diomedeinaa) : — ■

The more or less dorsal position of the nostrils, the form of which however varies, as
has already been described, though they are never lateral. The absence of a gluteus
quintus. The peculiar form of the biceps brachii muscle, which is in two separate
parts, the humeral head forming a patagial slip. The presence of haemapophyses on the
dorsal vertebra?, the centre of which are marked by more or less developed pneumatic
depressions. The non-pneumatic humerus. The different pterylosis, and the nearly
equal size of the lobes of the liver. The greater size of the hallux, which always has a
distinct nail externally. (Quite absent in Pelecanoldes.)

Pelecanoides is, in some respects, as much specialised as the Albatrosses, though many
■of its modifications are distinctly traceable to its diving habits, as, e.g., the compressed form
of the wing bones, the great development of the bypapophyses of the dorsal vertebrae,
the elongated sternum and pectoral muscles, the peculiar ribs. But it stands alone
(amongst the Procellariidae) in the absence of the ambiens muscle ; the peculiar disposi-
tion of the femoral vein ; the absence of a hallux ; and the single interclavicular air-cell.
Moreover, as in Bulweria only of other Tubinares, its myological formula is A.X., there
being no accessory head to the femoro-caudal muscle.

1 This feature, in which the Albatrosses are apparently more primitive than are either the Oceamtitlcc or the other
Procellariidae, can hardly, if my views about the relationships of these groups to each other be correct, be considered to
have been a character of the common Petrel-ancestor. It may be more probably explained as due to arrested develop-
ment during embryonic life, as a study of the development of the nostrils of other Petrels would probably show that
these are actually, at some time, lateral, anil subsequently coalesce

(ZOOL. CHALL. EXP\ PART XI. —1882.) L 8

58 THE VOYAGE OF H.M.S. CHALLENGER.

But Peleeanoides shows marks of being in some respects an early form in the simple
condition of the tensor patagii muscle, in its very simple syrinx, and in the general
shape of its sternum. It has the characteristic form of biceps muscle found in all the
Procellariidse, except the Albatrosses, and like all those forms, except the Procellaria-
group, has basipterygoid facets.

Peleeanoides is thus, as will be seen, a very well-marked form, though it is somewhat
difficult to decide as to whether its peculiarities are such as to entitle it to form a
separate sub-family by itself. The presence of basipterygoid facets would seem to
indicate that it probably diverged from the general stock of the Procellariinaj at a point
when the latter had already developed that feature, and therefore at a period after the
ancestor of the Procellaria-gvowp — in many ways the least specialised, and therefore pre-
sumably more ancient, of the sub-family, and in which there are no such facets — had
already acquired its main characters.

According to modern ideas, the object of a classification is not so much to represent
morphological facts as to indicate the phylogenetic relations of the different forms con-
cerned. According to the first view, Peleeanoides might well be placed, as many authors
have done, in a special group of its own ; but if we admit, as seems on the whole most
probable, that it has been derived from the same stock as the Procelko-ia-gtowp after the
special ancestor of the latter was developed, I prefer considering it as simply a highly-
specialised form of the Procellariime.

The Procellariinse so defined fall into a number of smaller groups, distinguishable by
good characters.

The "Stormy-Petrels" of the genera Procellaria, Cymochorea, and Halocyptena1
form one such minor group, distinguished by their general small size and coloration, com-
paratively long tarsi, nearly single nasal aperture, simple triangular tongue, simple
tensor patagii, peculiar skull wTith no basipterygoid facets or distinct uncinate bone,
entire posterior sternal margin, and little specialised syrinx. Procellaria has two
cseca, Cijmochorea one only, and Halocyptena, as already mentioned, has them quite
absent.

The position of Peleeanoides has already been fully discussed ; it stands quite per se,
though presumably derived from a stem common to it and the remaining Procellariinse,
which must have diverged from the less specialised one now represented by the
Proce^a n'rt -group.

Prion (with which JIalolxena is probably to be associated) represents a third minor
group, much specialised as regards its peculiarly broad beak with its fringe of lamella?,
whilst in its tensor pataejii arrangement and syrinx it is not highly developed.

The two genera Pagodroma and Daption seem very central as regards their relation-
ships, which seem to be with Prion (as indicated chiefly by the rudimentary lamella? of

1 Oceanodroina also, I have little doubt, belongs to this group.

REPORT ON THE ANATOMY OF THE PETRELS.

59

Daption) on the cue hand, and with the Fulmars on the other, Aeipetes* being the
less specialised of these, both as regards its imperfect tracheal septum, and the number
of rectrices. The type of syrinx so characteristic of the Fulmars is foreshadowed, a.s it
were, as has been already pointed out (supra, p. 35) in that of Pagodroma, and all
four genera (Fulmarus, Thalassceca, Ossifraga, and Aeipetes) agree in the general
disposition of the tensor patagii, which has no ossicles, in the more or less rudimentary
os uncinatum in the tendency to anchylosis of the lachrymal and frontal, in the shape
of the tongue and of more or less well-developed lamellae on the bill, and in having four
more or less complete, but never deep, sternal emarginations.

Aeipetes is, on the whole, the least specialised of the Fulmarine group in the most
limited sense. This includes besides Thalassceca, Fulmarus, and Ossifraga, which last,

a,

s&

Fig. 32.

Fig. 31. — Beak of TJialassaca glaciahidcs. a. The aperture of the nasal tubes,

from in front. Natural size.

Fio. 32. — The same parts of Aeipetes antarcticus.

on account of its great size, peculiar syrinx, and sixteen rectrices, may be considered the
culminating point in this direction of the Procellariidae.

The remaining genera, CEstrelata, Puffinus, Adamastor, Majaqueus, and Bulweria
are also apparently closely related to each other, the first and last named being perhaps

1 I propose to make a germs under this name, for the reception of the Proeellaria antarctica of Gmelin (Syst. Nat.,
17S8, vol. i. p. 565), which has usually been considered congeneric with Thalassceca, the type (and only representative)
of which is Thalassceca glacialoi3.es. For the latter bird also was instituted Hombron and Jacquinot's genus Priocella
(s.c, vol. iii. p. 148). Aeipetes is easily distinguishable from Thalassceca by the much shorter and stouter bill, and
differently shaped nasal tubes, as will be best understood from the accompanying figures (figs. 31, 32). The number
of rectrices is also different (twelve as compared to fourteen) ; the tracheal septum is incomplete, and the structure of
the syrinx also quite different {vide supra, p. 37). The coloration of the two forms is quite unlike.

(50 THE VOYAGE OF H.M.S. CHALLENGER.

least so. All agree in having a deeply four-notched sternum, in having well developed
uncinate bones, in the possession of one or two accessory wing-ossicles developed in the
termination of the tensor patagii tendons, in the spiny tongue (? Adamastor), and the
palatal armature of spines (? Adcimastor), and in there never being even indications of
lateral lamellaj on the beak.

CEstrelata differs from its allies in having only a single ulnar ossicle, there being two
in all the others.

Puffinus ami Adamastor are more closely connected together than they are with
Majaqueus, easily distinguishable by its more normal nostrils, less compressed tarsi, and
specialised (? Adamastor) syrinx. Bulweria is a peculiar form, with no very close ally,
and must be regarded as a highly specialised form, as shown in its myological formula
being reduced to A.X, and its peculiar cuneate tail. It has no close relationship at all
to the Stormy-Petrels, as already pointed out by Dr. Coues,1 and Garrod.2

These views on the classification of the Tubinares may be represented in the annexed
diagram (p. Gl).

V. THE AFFINITIES OF THE TUBINARES.

The Tubinares as a group may be shortly defined as follows : —

Holorhinal schizognathous birds with a large, broad, depressed, pointed vomer, and
truncated mandible ; with the anterior toes fully webbed, and the hallux either very small
and reduced to one phalanx, or absent ; with a tufted oil gland and large supra-orbital glands
furrowing the skull ; with the external nostrils produced into tubes, usually more or less
united together dorsally ; with an enormous glandular proventriculus and small gizzard
of unusual shape and position, and with the commencing duodenum ascending ; with a
completely double great pectoral muscle, and a well-developed pectoralis tertius ; with
the femoro-caudal and semi-tendinoms muscles always present, and the ambiens and
accessory femoro-caudal only exceptionally absent.

Some, at least of these characters — the structure of the hallux, the formation of the
nostrils3 and the form of the stomach are cpfite peculiar to the Tubinares, not being
found in any other birds, though of universal presence in these. These features alone
would at once suffice to distinguish them from any other Avian order, whilst the combin-
ation of other characters is as unique. It is therefore a difficult task to assign to this
group a satisfactory position in any arrangement of the class Aves, owing to its much
isolated position.

S.C., 18C6, p. 13!). - Coll. Papers, p. 221.

3 The Caprimulgine genus Siphonorhis (Sclater, Proc. Zool. Soc, 1861, p. 78) perhaps approaches the Tubinares

more nearly in this point than any other hint known to me.

<}'2 THE VOYAGE OF H.M.S. CHALLENGER.

Most previous writers have considered the Petrels as more or less closely connected
with the Gulls (Laridae), hut the grounds for any such collocation are very slight, in my
judgment, now that the structure of the two groups is better known.

The Gulls exhibit no trace of any of the characteristic peculiarities of the Petrels,1 and
differ widely from them in the important feature of being schizorhinal.2 The peculiar
disposition in two quite separate layers of the great pectoral muscle in the Tubinares is
quite unlike anything seen in the Gulls or their allies, whilst the large pectoralis
tertius of the Petrels is altogether unrepresented in the Laridaa. The character of
the caeca in the two groups is also quite different, and there are no special osteological
resemblances between the two groups so far as I can see, for the mere schizognathous
character of the palate is, we now know, not necessarily a mark of affinity. The
character of the young plumage, the condition of the young birds, and the number,
shape, and coloration of the eggs — points on which some stress may be laid in questions
of this kind — are totally dissimilar in the two groups, as indeed are the habits of the
adult birds themselves, though no doubt both are "web-footed" and more or less
pelagic in habit. Such resemblances, however, can* hardly be seriously considered as
indicating any real affinities.3

L'herminier, A. Milne-Edwards, and Huxley have all, in describing various points in
the osteology of the Tubinares, pointed out similarities of various kinds between their
osseous structure and that of various forms of the Steganopodes, though they still kept
them close to the Laridae. Eyton, on the other hand, places the various Petrels he
describes in the family " Pelecanidae," the Gulls forming a separate family by them-
selves.

But no one will be prepared, I think, to dispute that the Steganopodes are allied to
the Herodiones, including under that name the Storks and Herons, with Scopms, only.
Thus, on osteological grounds alone, there is sufficient ground for placing the Tubinares
in the vicinity of the Steganopodes and Herodiones. And, in fact, neglecting the
desmognathous structure of the palate — the taxonomic value of which per se is becoming
more and more dubious as our knowledge of the structure of birds increases — there is little
in the characters assigned to the groups Pelargomorphas and Dysporomorphae by Professor
Huxley (I. c, p. 461) that is not applicable to the general Petrel type.

The completely double great pectoral muscle is a characteristic only found, as already
•observed, in the Ciconiidaa, Cathartidse, the Steganopodes (except Phahierocorax), and

1 I cannot understand Professor Huxley's remark (Proc. Zool. Soc, 1867, p. 455) that " the Gulls grade insensibly
into the Procellariid;e."

• '/. Garrod, Coll. Papers, p. 128.

3 No views regarding the affinities of the Petrels other than that to the Laridae already discussed, and that to
the Ciconiiform birds have, so far as I know, been seriously advanced by ornithological writers, Professor Garrod
having abandoned his early idea that the Tubinares were probably related remotely to the Anseres and their allies,
(r/. Coll. Papers, pp. 220 and 521).

REPORT ON THE ANATOMY OF THE PETRELS. 63

the Tubinarcs, and in all these forms it is associated with short colic caeca of peculiar
shape (absent altogether in the Cathartidse, as in some of the Tubinarcs), more or less
completely webbed feet, tufted oil-gland (except in the Cathartidge), holorhinal nostrils,
a tendency of the palatine bones to unite behind the posterior nares, truncated mandible,
broad, strong, well-developed sternum, and strongly curved, well-developed clavicles.
These birds also agree together in being " Altrices," the young birds being quite helpless
after birth, and requiring to be fed for a long time by their parents — and in generally
laying eggs of a white, or nearly white, colour.

The group so constituted, of which the Ardeidge and Falconidge must also be considered
as aberrant members, — the first family being closely related to the Ciconiidge through
Scojnis, whilst the Falconidge are probably, though much more remotely, connected with
the Steganopods, — corresponds to the Ciconiiformes of Garrod,1 with the addition, as he
had already himself suggested,2 of the Tubinares.

But his earlier definition of that group, in so far as it relates to the absence in it of
the accessory femoro-caudal muscle (B), will have to be modified, inasmuch as this
muscle is, as shown above, generally present in the Tubinares. These too, differ markedly
from the other Ciconiiformes in the well -developed pectoralis tertius (very small or
absent in the others), in the large size of the vomer, and the non-desmognathism of the
palate, though as regards this latter character it has already been pointed out that the
Albatrosses are nearly desmognathous, whilst the desmognathism of the Cathartidge is
of a different kind to that prevalent in the other forms concerned.

The two existing groups of Petrels are clearly related to each other so much more nearly
than to any other group of birds that it is evident that they must have had a common
ancestor that possessed the peculiar features characterising the Tubinares as an order.
Such a form may therefore be safely assumed to have had —

1. The characteristic nostrils of the group.

2. The equally characteristic stomach and duodenum.

3. Webbed feet, with a small hallux of a single phalanx.

4. A double great pectoral muscle, and large jwctoralis tertius.

5. A formula AB.XY, a gluteus j^rimus and an ambiens muscle.

6. Short colic cgeca of characteristic shape.

7. A tufted oil gland, and the pterylosis characteristic of the group.

8. A holorhinal schizognathous skull, with large depressed vomer, great supra-orbital
glandular depressions, no basipterygoid facets, and a truncated mandible.

9. A short, broad, deeply-keeled sternum, more or less entire behind, with strong
clavicles.

10. A peculiar humerus, and tibia with large cnemial crest.

1 Collected Papers, p. 213. - Lor. ci'., p. 521.

C4 THE VOYAGE OF H.M.S. CHALLENGER.

No living Petrel has this combination of characters ; the Oceanitidse having lost their
colic caeca, the Procellariidas the accessory semi-temlinosus (Y) muscle, and both groups
having become specialised in other ways.

Such an ancestral form as here indicated may be supposed to be an early, and in some
respects — as shown by the large vomer, schizognathous palate, large third pectoral
muscle and formula AB. XY — more primitive form, that diverged from the common stock
of the Ciconiiform birds very early, when the latter had only acquired the most prevalent
of the characters now existing in the various groups of that sub- order. One branch of
this stock has since become greatly modified in the Tubinarial direction, whilst the other
branch, loosing " B " and the large vomer, and becoming desmognathous, split up and gave
origin, at different times and in different ways, to the remaining families of the group.
The definiteness of the characters of these, and the amount of specialisation they
show, indicate not only a great antiquity for the whole group, but also the great amount
of extinction that has gone on amongst its members in the past, in the process of which
nearly all the intermediate and less specialised forms have disappeared.

PLATE I.

Fig. 1. Head of Oceanites oceanicus. la leg and lb foot (from before) of same.

Fig. 2, 2a, lb. The same of Garrodia nereis.

Fig. 3, 3a, 36. The same of Pelagodrorna marina.

Fig. 4, 4a, 46. The same of Fregetta grallaria.

Fig. 5, 5a, 5b. The same of Procellaria pelagica.

Fie. 6. Bill of Bulweria columbina.

*&'

Fig. 7. View of external nares of same, from before, to show their two distinct
openings.

Fig. 8. Nostrils of Oceanites oceanicus, from before.

Fig. 9. The same of Procellaria pelagica.

Figs. 7-9 enlarged. The others are of the natural size.

The Voyacje of II M

Anaiomj of Petrels PI ]

^

^*

f

J. Sm.

EXTERNAL CHARACTERS OF PETRELS
(Oceanitidae,<fcc.)

PLATE II.

Fig. 1. Liver, stomach, and duodenal loop o£ Majaqueus cequinoctialis, viewed from in
front. L. Liver, pr. Proventriculus (the letters are placed on its fundus), g. Gizzard.
p. Pancreas, g.b. Gall-bladder, r.h.d., l.h.d. Right and left hepatic ducts, v.p. Vena
portae. h.a. Hepatic artery.

Fig. 2. Outline of stomach, &c, of Fregetta grallaria.

Fig. 3. Colic caeca of Majaqueus cequinoctialis, enlarged slightly.

Fig. 4. Gizzard of Fulmarus glacialis laid open, to show the character of its
epithelium.

Fig. 5. Tongue of Oceanites oceanicus.

Fig. 6. Tongue of Pelecanoides urinatrix.

Fig. 7. Tongue of Diomedea brachyura.

Fig. 8. Tongue of Phcebetria fuliginosa.

Fig. 9. Tongue of Oymochorea leucorrhoa.

Fig. 10. Tongue of Ossifraga gigantea.

Fig. 11. Tongue of Aeipetes an tarcticus.

Fig. 12. Tongue of Daption capensis.

Fig. 13. Tongue of Prion banksi.

Fig. 14. Tongue of Pagodroma nivea.

Fig. 15. Tongue of CEstrelata lessoni.

Fig. 16. Tongue of Larus sp. ?

Fig. 17. Tongue of Majaqueus cequinoctialis.

Fig. 18. Tongue of Puffinus brevicauda.

Fig. 19. Palate of Oceanites oceanicus, enlarged. (The line shows the natural size.)

Fig. 20. Palate of Pelecanoides urinatrix.

Fig. 21. Palate of CEstrelata lessoni.

Fig. 22. Palate of Fulmarus glacialis.

Fig. 23. Palate of Prion banksi.

All the figures, except figs. 3 and 19, axe of the natural size.

The VI

Lpels I

14-

jh

o

I

«

m#

!

76

20

«U|

T

r=i

ALIMENTARY CAN A L, TO N C U E, & PALATE OF PETRELS

PLATE III.

Fig. 1. Left pectoral region of Majaqueus cequinoctialis, to show the double
pectoralis major {sett primus) muscle. Its superficial layer (p. la) has been for the
most part removed, its cut origin from the sternal crest and furcula (F.) being reflected :
p. la', its insertion into the humerus, also cut and reflected, p. lb. The deep layer, p. 2.
Fascia covering the 'pectoralis secundus muscle, t.p. Belly of tensor patagii muscles.
t.p.l. Their tendon, joined by b.s. (biceps-sli'p) formed by the humeral head of the
biceps humeri muscle, b. Main belly of biceps muscle, formed by the coracoidal
head. S. Body of sternum, bare of muscular fibres.

Fig. 2. Dissection of left pectoral region of Diomedea brachyura to show the two
layers of the pectoralis primus muscle (p. la, p. lb), which have been cut and removed
in large part, and the compound pectoralis secundus. p. 2. Its sternal origin ; p. 2', its
coracoid origin ; p. 2", its furcular origin ; p. 2'", origin from coraco-furcular membrane
(c.f.m). p. 3. Pectoralis tertius. C. Coracoid bone. F. Furcula, at symphysis. S.
Sternum. (The line above it shows the limit of origin of the deep layer of the
pectoralis primus.)

Fig. 3. Dissection of right wing of Oceanites oceanicus, to show the peculiar
expansor secundariorum muscle, e.s. Tendinous portion arising from the last remiges
(S.). e.s'. The other moiety, arising from the last scapular feathers (Sc). p. 1.
Pectoralis primus muscle, to which the expansor secundariorum is attached. H.
Humerus. Pat. Patagial membrane, v.n. Vessels and nerves to wing.

Fig. 4. Left shoulder joint, inner side, of Ossifraga gigantea, to show the peculiar
biceps muscle, c. Its coracoid head, continuous below with the belly of the muscle
(b). h. Its humeral head, which forms a biceps-sli^, joining the tendon of the
tensor patagii (t.p). p. 1. Insertions (cut) of the two layers of the pectoral's
primus muscle, e. Extensor, t. Teres, c.b.l. Coraco-brachialis longus. c.b.b. Coraco-
brachial brevis.

Fig. 5. Same parts in Thalassiarche culminata. c.,h. Coracoid and humeral heads
of biceps, here uniting below into the common tendon of that muscle (b.) b.s.
Biceps slip, largely tendinous and joining the tensor patagii near the elbow, derived
from the coracoid head of the biceps.

,1

The Voyage ' Vnqer

And n Petrels P

P i<y

ebb

1 SmitlilK

MYOLOCY OF PETRELS.
Anterior Extremity

PLATE IV.

Fig. 1. Dissection of right elbow of Prion desolatus, to show the disposition of the
tensor patagii muscles, as seen from above.

Fig. 2. The same in (Estrelata brevirostis.

Fig. 3. The same in Diomedea exulans. An arrow is passed between the twin
tendons of origin of the superficial part of the extensor metacarpi radialis longior.

Fig. 4. The same in (Estrelata lessoni.

Fig. 5. The same in Ossifraga gigantea.

Fig. 6. The same in Pelecanoides urinatrix. H. Humerus. R. Radius.
b. Biceps muscle, t.p.l. Tensor patagii longus. t.p.b. Tensor patagia brevis.
e.m,e.m'. Superficial and deep bellies of extensor metacarpi radialis longior. e.m*
Inner of twin tendons of origin of its superficial belly, b.s. (in fig. 3). Biceps slip.
f. Fasciculus of patagial tendons continued on to ulnar fascia. a,a'. Ossicles developed
at origin of the extensor metacarpi radialis longior. t.p.' (in fig. 5). Special slip from
patagial tendons to deep belly of extensor metacarpi 7'adialis longior.

Fig. 7. Dissection of right wing of Majaqueus csquinoctialis, to show origin and
general disposition of the tensor patagii muscles. Lettering as above ; also t.p.
Common belly of tensor patagii longus and brevis. t.p.l'. Cushion of elastic tissue,
developed in the tendon of the tensor patagii longus (t.p.l.) at its origin from the
humerus, e. Elastic pad, developed in the marginal tendon of tensor patagii longus,
opposite the elbow. d. Deltoid muscle. l.d. Latissimusdorsi (insertion), n. Cir-
cumflex nerve.

•Voyaae of H M S Challenger

Anatomy of Petrels P] IV

MYOLOGY OF PETRELS
I i.aii musi les

PLATE V.

Fig. 1. View of superficial muscles of right thigh of Majaqueus cequinoctialis.
P. Pelvis, s. Sartorius. gl. 1. Gluteus primus, g. Gastrocnemius, b. Biceps, s.t.
Semi-tendinosus. s.m. Semi-membranosus.

Fig. 2. View of deeper thigh muscles of the same bird ; the gluteus primus, biceps,
and gastrocnemius (</.) muscles cut and reflected to show the deeper parts. Lettering
as above ; also b'. cut end of biceps passing through the tendinous loop formed by
the origin of the gastrocnemius, f.c. Femoro-caudal muscle, a.f.c. Its accessory head.
o.e. Obturator externus. add. Adductor muscles, f.v. Femoral vein, sc. Sciatic nerve
and artery, o.g. Oil gland.

Fig. 3. The same parts in Oceanites oceanicus. a.s.t. Accessory semi-tendinosus
muscle. Pb. Pubis. R. Rectrices.

Fig. 4. Dissection of thigh of Pelecanoides, to show the absence of the accessory
femoro-caudal muscle, and the abnormal course of the femoral vein, this passing over,
instead of under, the femoro-caudal muscle, o.i. Obturator internus. gl. Glutei.

The Voyage of HM.S'Ch . |

Anatomy of Petrel.': .

J Smii IiU\

MYOLOCY OF PETRELS
Hinder Extremity

. ft :mo

PLATE VI.

Fig. 1. Skull of CEstrclata lessoni, viewed from the side. All the figures are of the
natural size.

Fig. 2. The same, from below.

Fig. 3. The same, from above.

Fig. 4. Skull of Prion vittatus, from below.

Fig. 5. Vomer, — with the ascending plate, anchylosed to it, of the palatine bones,
— of Diomedea exulans, from the side.

Fig. 6. The same, from above.

Fig. 7. Left uncinate bone (" os crochu " of Reinhardt) from behind, of Thalassiarche
culminata.

Fig. 8. The same of Phcebetria fuliginosa.

Fig. 9. Left humerus of Majaqueus cequinoctialis, from above.

Fig. 10. Proximal, and fig. 11, distal extremities of the same bone, inferior surface.

Fig. 12. Pelvis, seen from the side, of Majaqueus cequinoctialis.

Fig. 13. Proximal end of right tibia of Ossifraga gigantea, to show the cnemial crest.

Fig. 14. Hallux of Ossifraga, with its metatarsal. The single phalanx of which it
consists is vertically bisected, to show the hollow interior.

i -nger

OSTEOLOGY OF PETRELS

Ski, ,

\wp.

PLATE VII.

Fig. 1. Sternum, with coracoids and furcula, of Thalassiarche melanophrys , from
in front. Reduced.

Fig. 2. The same, from the side.

Fig. 3. Sternum and pectoral arch of Pelecanoides urinatrix, from in front.

Fig. 4. The same, from the side.

Fig. 5. The same, of Cymochorea leucorrhoa, from in front.

Fig. 6. The same, from the side.

Fig. 7. The same of Fregetta rnelanogastra, from in front.

Fig. 8. The same, from the side.

Fig. 9. Outline of posterior margin (right side) of sternum of Ossifraga gigantea.
Reduced.

Fig. 10, The same of Da<ption capensis.

Fig. 11. The same of Fulmar us glacialis.

Fig. 12. The same of Aeipetes antarcticus.

Fig. 13. The same of Thalassoeca glacialoides.

Fig. 14. The same of Prion desolatus.

Fig. 15. The same of Pagodroma nivea.

Fig. 16. The same of Bidweria columbina.

Fig. 17. The same of GLstrelata lessoni.

Fig. 18. The same of A damastor cinereus.

Fig. 19. The same of Majaqueus cequinoctialis.

Fig. 20. The same of Puffinus anglorum.

•

I i

10

11

14-

15

J Smitlith

1

w

16

n

77

r

OSTEOLOGY OF PETRELS

/8

\ f

\)
13

Hftr.hai

THE

VOYAGE OF H.M.S. CHALLENGED

ZOOLOGY.

REPORT on the Deep-sea Medusa dredged by H.M.S. Challenger during the
years 1873-1876. By Prof. Ernst PLeckel.

PREFACE.

The Deep-sea Medus/E form one of the smallest and least important groups of the
rich and remarkable deep-sea fauna for whose discovery we are indebted to the voyage of
H.M.S. Challenger; the number of the species described does not exceed eighteen, of
which half are Craspedotas and half Acraspedse. The majority of these eighteen species,
however, which represent an equal number of genera, are of great morphological interest,
and they, moreover, claim special attention as the first deep-sea inhabitants of this class
of animals with which we have become acquainted. I am therefore particularly in-
debted to Sir Wyville Thomson for most liberally handing over to me for examination
the whole of the Medusas collected by the expedition. To the first large collection of
deep-sea Medusae, sent to me in 1877, I was able to add a few more, which I found
incidentally in examining the Radiolaria of the expedition, and in going more minutely
over the general collection, during a visit to Edinburgh in 1879. I have especially to
thank Mr. John Murray, principal assistant on the civilian staff of the Challenger, for
kindly aiding me in this work in every way in his power.

All the eighteen species of deep-sea Medusae described in this Memoir have already
received a brief diagnosis in my System der Medusen published in 1879. Besides
these, I had already described two other species, the iEginid, sEginorhodus rosarius
(System, p. 345, No. 379), and the Cyaneid, Melusina formosa (System, p. 535, No. 500).

2. (ZOOL. CHALL. EXP. PART XII. 1881.) M a

ii THE VOYAGE OF H.M.S. CHALLENGER.

Unfortunately, however, the fragments of the only specimens which I obtained of these
two beautiful Medusae, proved on closer examination to be so imperfect and so badly pre-
served, that I was obliged to give up all idea of their further description and illustration.
On the other hand, I gained a special advantage in the description of a few other deep-
sea Medusae, whose single specimens from the Challenger collection were also insufficiently
preserved, as I was able to complete it by means of material procured elsewhere.
I have to thank Professor Steenstrup and Dr. Lutken for several well-preserved specimens
of Ptychogena pinnvXata and Pectyllis arctica from the Zoological Museum of Copen-
hagen, and my friend Mr. Gregor Buccich for several excellent specimens of Drymonema
victoria from the island of Lesina in Dalmatia. Three species, of each of which the Chal-
lenger collection only contained an imperfect fragment, I took myself living with the
tow-net, and thus had an opportunity of examining them minutely ; Pectanthis aster-
oides from the Mediterranean, Cunarcha ceginoides from Lanzarote, one of the Canary
Islands, and Polycolpa forskalii from the Red Sea. Of the majority of the twelve other
species I had, unfortunately, only a single specimen at my disposal, and that was often
imperfect; still I hope that, with the assistance of the comparative morphology of the
most closely allied Medusae, I have satisfactorily reconstructed their organisation.

It is by no means certain that all the eighteen deep-sea Medusae described below are
constant inhabitants of the deep sea ; the method of capture by the tow-net by which
such delicate and fragile organisms are often brought from great depths of the sea is still
imperfect, and it is probable that the greater number of Medusae brought up apparently
from the greater depths, really swim in shallower water and are only taken in " hauling
in " the net. Those Medusae, on the other hand, may be regarded with greater proba-
bility as permanent and characteristic inhabitants of the deep sea, which have either
adapted themselves by special modifications of organisation to such a mode of life, or which
give evidence by then- primitive structure of a remote phylogenetic origin. As such I
consider of the nine Craspedotae described, — first of all the three Pectyllidae (Pectyllis,
Pedis, and Pectanthis), and secondly Cunarcha and jEginura; among the nine Acra-
spedae, Tesserantha, the two large Periphyllidae (Periphylla and Periphema), and the
two remarkable Ephyridae (Nauphanta and Atolla). It appears, however, from the occa-
sional observations of single naturalists, that other Medusae species also (especially
Charybdeidae and Pdiizostomidse) inhabit for the most part the bottom of the sea, and
pass into considerable depths. We may therefore hope that the following description of

REPORT ON THE DEEr-SEA MEDUSAE. iii

the deep-sea Medusae may be regarded as the first instalment from an interesting region
from which we may expect many and important conclusions as to the organisation of
the Medusae.

It will be seen from the systematic survey on p. 141 that altogether eight orders of
this class, distinguished by me in my System der Medusen, 1879, are represented
among the eighteen deep-sea Medusae of the Challenger expedition. Of these one species
(Thamnostylus, PI. I.) belongs to Order I. Anthomedusge ; one species [Ptyclwxjena,
PI. II.) to Order II. Leptomedusae ; three species (Pectyllis, Pedis, and Pectanthis,
Pis. III.— VIII.) to Order III. Trachomedusae ; four species (Cunarcha, Polycolpa,
Pegantha, JEgiimra, Pis. IX.-XIV.), to Order IV. Narcomedusae ; two species (Tessev-
antha and Lucemaria, Pis. XV.-XVII.) to Order V. Staurontedusae ; two species (Peri-
phylla and Periphema, Pis. XVIII.-XXV.) to Order VI. Peromedusaa ; one species
(CJiarybdea, PI. XXVI.) to Order VII. Cubomedusae ; four species (Nauphanta, Atolla,
Drymoncma, Leonura, Pis. XXVII.-XXXII.) to Order VIII. Discomedusae. Thus
thirteen of the thirty -two families which I defined in my System der Medusen, 1879, are
represented here (comp. p. 141).

A tabular view of the chorology of these Medusae is given on p. 142. Their
geographical distribution extends throughout the whole of the great oceans; on the whole,
there are eight species in the Atlantic-Mediterranean region, and ten species in the Indo-
Pacific region. Of the eight former species seven belong to the northern, one to the
southern half of the Atlantic Ocean (two of which are also found in the Mediterranean).
Of the remaining ten species, two belong to the north half, three to the southern half
of the Pacific Ocean, and five to the Antarctic part of the Indian Ocean. One of the
latter is also found in the south-western part of the Atlantic.

With regard to the bathymetrical distribution, it will be seen from the table on p. 142
that seven species were taken in depths from 80 to 600 fathoms, six species in depths
from 1100 to 1600 fathoms, and five species in depths from 2000 to 2200 fathoms. For
reasons already stated, these depths must (partly at least) be considered more or less
approximate.

With regard to the figures of the deep-sea Medusas in the' thirty-two plates appended
to this Memoir, it is of course impossible, from the imperfect state of preservation of the
spirit specimens, to expect that they should be absolutely true to nature. I rather con-
sidered it my duty here, as in those figures in my System der Medusen which were

iv THE VOYAGE OF H.M.S. CHALLENGER.

drawn from spirit specimens, to take advantage of my knowledge of the forms of the
living Medusae, to reconstruct the most probable approximate image of the living forms.
I was greatly assisted in my efforts in this direction by the skilful hand of my litho-
grapher, Mr. Adolf Giltsch. I am also greatly indebted to my friend Dr Keinhold
Teuscher for preparing a large number of excellent microscopic sections. This Memoir
has been translated into English by Miss Nellie Maclagan, and I thank her for the kind
care with which she has executed a difficult task.

The many new morphological facts, furnished by close examination of the deep-sea
Medusae, are not only of special interest in themselves, but are, for the most part, of
general interest for deciphering the comparative anatomy of the whole class ; this is true,
for example, of the Pectyllidse and Peganthidae among the Craspedotae, and of the
Periphyllida? and Ephyridae, among the Acraspedae. I therefore considered it convenient
to preface the special anatomical description of the eighteen deep-sea Medusae by a short
article on the Organisation of the Medusae, which serves in a certain measure as a mor-
phological introduction to the former, and at the same time as a preliminary Scheme
of a comparative Morphology of the Medusae.

EENST H^ECKEL.
Jena, lith September 1881.

REPORT ON THE DEEP-SEA MEDUSAE.

THE ORGANISATION OF THE MEDUSA.

Sketch of a Comparative Morphology of the Medusae as an Introduction to the
Description of the Deep-sea Medusa of the Challenger Expedition.

I. GENERAL MORPHOLOGY OF THE MEDUSiE.

§ 1. Definition of the Medusas. Medusas are Cnidarias or Acalephae with a gelatinous,
radially constructed, concave-convex umbrella, whose vertical axis is the principal axis
of the single persona ; with swimming muscles on the concave oral side of the umbrella ;
nerve centres and organs of sense on the peripheric margin of the umbrella ; with radial
processes (canals or pouches) of the central gastral cavity, and a simple (seldom multiple)
oral opening at the oral pole of the principal axis ; also with genitalia in the subumbral
wall of the gastrovascular system (comp. System der Medusen, 1879, p. 21, &c). Medusae
are distinguished from the other classes of Acalephae or Cnidarise (Polyps, Corals, Siphono-
phora, Ctenophora) by the following class characters : the mature, completely developed
Medusa appears as a single persona (not united in numbers in a colony), which usually
swims freely in the water, seldom crawls, still more rarely is fastened to the bottom ;
(nearly all Medusas live in the sea, only a few in fresh water). The principal mass of the
Medusa body (by volume and weight) forms a concave-convex, watery gelatinous body,
the umbrella. This serves as the special swimming organ ; a vertical principal axis and
two or more horizontal transverse axes (with twice as many radial axes) are always the
standard for its pyramidal base form. The convex outer surface of the umbrella (exum-
brella) does not generally bear special organs, whilst the concave inner surface (sub-
umbrella) is always overlaid with a muscular plate, which, as the most important organ of
motion, deepens and narrows the umbrella cavity by its contraction, and thereby expels the
water from it. The most important organs of sensation, the nerve centres, and differenti ■
ated organs of sense, and generally the tentacles also, lie on the umbrella margin (where
the exumbrella and the subumbrella are contiguous). The organs of nutrition are
formed by a radial gastrovascular system which extends over the subumbrella, and is
composed of a central principal intestine and a peripheric coronal intestine. The principal
intestine (in the centre of the subumbrella) forms a simple gastral cavity, which often pro-
jects below like a tube and opens at the oral pole of the principal axis by a simple (seldom
multiple) mouth. The coronal intestine (in the periphery of the subumbrella) is composed
of four or more radial processes of the central stomach, which sometimes take the form of
broad pouches, sometimes of narrow canals. The organs of reproduction are always simple
reproductive glands (genitalia) and are always developed in the lower (subumbral)
wall of the gastrovascular system, sometimes from its endoderm, sometimes from its

vi THE VOYAGE OF H.M.S. CHALLENGER.

ectoderm. Nearly all Medusae are gonochoristie, only a few hermaphrodite. As regards
the phylogeny of the Medusae, their comparative anatomy and ontogeny points out the
class of Polyps as their original ancestral group.

§ 2. Definition of the Polyp. The class of the Polyps, from which the class of Medusa?
are phylogenetically derived, and which, therefore, furnishes the key to our morphological
knowledge of them, is at the same time the common ancestral group of all Cnidariae.
At present the fresh- water Polyp (Hydra), and the closely allied genera Clava, Coryne, &c.,
must be considered their simplest and most primitive representatives. Their developed
organism is immediately allied to that of Gastrula (phylogenetically Gastraea), and has the
same simple primitive intestine with primitive mouth, whose wall is formed from the two
primary germinal layers. The Polyp is chiefly distinguished from the Gastrula by begin-
ning histological differentiation of the two primary germinal layers, and also by the fact
that the aboral pole of the principal axis serves as a peduncle for adhesion, whilst a corona
of feelers or tentacles is developed round the mouth at its oral pole. This corona of ten-
tacles forms the boundary between the strongly-arched aboral part of the body(cup, "calyx")
and the flat or even depressed oral part (oral disk, " peristomium ").

§ 3. Medusa and Polyp. The most essential difference between the organisation of the
Medusa and the Polyp consists in the formation of the characterstic swimming organ of
the former, the umbrella, and of the cathamma or partial fusions between the aboral wall
of the cup (" calyx ") and the oral disk (" peristomium "), by means of which the peripheric
part of the simple gastral cavity is divided into radial pouches or canals, which regularly
surround its simple central space. The central gastral cavity of the Medusa does not
therefore correspond to the whole simple gastral space of the Polyps (or to the primitive
intestine of the Gastraea) but only to the central part of it, whilst on the other hand its
peripheric part is homologous with the radial (usually four-rayed) coronal intestine
of the Medusa (" pouch corona or canal corona "). The central oral opening is identical
in both animal classes, arising from the primitive mouth of the Gastrula, and when it is
prolonged in the Polyps into an oral cone or proboscis, the latter corresponds to the
freely projecting oesophagus or oral peduncle of many Medusa?. The peristomium of the
Polyps (or the slightly concave oral disk) is homologous with the " subumbrella " of
the Medusa (or the more strongly depressed lower surface of the umbrella). In the same
way the convex outer surface or the dorsal cup-wall (calyx) of the Polyp corresponds to
the more depressed dorsal wall or " exumbrella " of the Medusa. The cup-margin of the
Polyps (with the insertion of the corona of tentacles) is therefore homologous with the
umbrella margin of the Medusa, but in the latter, differentiated organs of sense are
developed beside or from the tentacles, whilst this is not the case in the former. As
regards histological differentiation, the older and lower Polyp form remains far behind
the younger and higher Medusa form. The latter has arisen from the former by adaptation
to a swimming mode of life, and has thereby become perfected.

REPORT ON THE DEEP-SEA MEDUSAE.

vn

§ 4. Craspedotae and Acraspedae (System, p. xxiv. 1, 3G1). The class of the Medusae is
divided into two different sections or sub-classes, Craspedotse and Acraspedae. Both sub-
classes form natural principal groups, differing essentially and thoroughly from one
another in important conditions of organisation. Although single groups in both sections
are so like that they may be confounded together, and, in fact, have often been so, yet
the two principal groups differ fundamentally, and have probably arisen cmite independ-
ently of one another from two different groups of polyps ; their most important
differences are best seen from the following tables: —

I. CRASPEDOTiB 01' HyDROMEDUSJE.

A. Gastral space, without gastral filaments or pha-

celli (Aphaeelhe).

B. Genitalia ectodermal (sexual products of the outer

germinal layer) (Cryptocarpa?).

C. Umbrella margin with true velum, without true

marginal lobes (Craspedotae).

D. Organs of sense usually simple, without protec-

tive plate (Gymnophthalmse).

E. Marginal nerve ring double and centralised

(Cycloneurce).

F. Descent from Hydropolyps, or Polyps without

gastral tseniola (Hydroniedusse).

II. ACRASPED.E or ScYPHOMEDUSiE.

A. Gastral space with gastral filaments or phacelli

(Phacellotse).

B. Genitalia endodermal (sexual products of the

inner germinal layer) (Phanerocarpre).

C. Umbrella margin without true velum, with true

marginal lobes (Acraspedae).

D. Organs of sense usually composite, with special

protective plate (Steganophthalmae).

E. Marginal nerve ring simple, usually non-central-

ised (Toponeurae).

F. Descent from Scyphopolyps, or Polyps with gas-

tral taeniola (Scyphomedusae).

§ 5. Hydropolyps and Scyphopolyps. The two sections of the class Medusae corre-
spond to two different sections of the Poly]) class, Hydropolyps and Scyphopolyps. The
lower more simply constructed Hydropolyps have retained a perfectly simple gastral
space with smooth inner surface (primitive intestine or "archigaster" of the Gastrcea). The
higher and more perfect Scyphopolyps are distinguished from the Hydropolyps by four in-
terradial longitudinal swellings or gastral ridges ("taeniola") being developed on the inner
surface of the gastral cavity, by which their peripheric part is consequently divided into
four broad perradial niches or pouches (" antra gastralia "). The Hydromedusae or
Craspedotae, as well as the classes of Siphonophorae, and Ctenophorae derived from them,
are descended from the Hydropolyps (Hydra, Clava, &c.) ; the gastral filaments are want-

viii THE VOYAGE OF H.M.S. CHALLENGER.

ing in all these Cnidariae. The Scyphomedusae or Acraspedae, as well as the class of Corals
or Anthozoa are descended from the Scyphopolyps (Scyphostoma, Spongicola, &c) ; both
these classes have gastral filaments or mesenteric filaments, which have arisen from the
taeniola of the Scyphopolyps.

§ 6. Polyphyletic origin of the Medusae. That the class of Medusas belongs to the
polyphyletic classes of animals is now a phylogenetic hypothesis, which may be brought
forward as a probability bordering upon certainty, although, on the one hand, the character-
istic structure of the Medusae appears so uniquely organised that they are most suitably
placed as a separate class in the system of the animal kingdom, yet, on the other hand, it
by no means follows that they are all derived from a single common ancestral form, which
already possessed the form of the Medusae. It is much more probable that the two sub-
classes, or sections of this class, the Craspedotae and Acraspedae, are of separate origin, and
are descended from groups of Polyps, which have developed into Medusae, independently
of one another. A strong support to this hypothesis is, that the Scyphopolyps, the ancestral
form of the Acraspedae (Scyphostoma, Stephanoscyphus, &c), already possess the four im-
portant interradial taeniola or gastral longitudinal ridges from which the four character-
istic groups of filaments are developed in all Acraspedae (or Scyphomedusae). On the
other hand, the characteristic groups of filaments are wanting in the Craspedotae (or
Hydromedusae) as the typical four interradial taeniola are wanting in their ancestral
form, the Hydropolyps. Moreover, the reproductive organs originate in the Craspedotae
(as in most Hydropolyps) from the ectoderm and in the Acraspedae (as in the Scyphopolyps
and Corals) from the endoderm. As regards the two sections or sub-classes, the Craspedotae
are more probably of monophyletic origin, the Acraspedae of polyphyletic.

§ 7. Orders of the Craspedotae (System, pp. 2, 233, 360). The section of the
Craspedotae or Hydromedusae is divided into two sub-sections and four orders. The two
sub-sections, Leptolinae and Trachylinae, are thoroughly and pre-eminently distinguished
from one another by the absence or presence of the cordyli or tentacular "auditory clubs."
These are modified acoustic tentacles, consisting of a solid axis of chordal endodermal cells, of
which the last (distal) contains one or more otolites ; their ectodermal epithelium bears stiff
auditory bristles. The first sub-section (or Acordyliae) has no auditory clubs ; it has, in
fact, either no auditory organs or only " velar auditory vesicles" (marginal vesicles on the
velum with ectodermal otolites) which are quite different from the cordyli, and occur in no
other group. Moreover in the Leptolinae, the tentacles are usually hollow, very movable
contractile filaments, and their velum is delicate and thin. The second sub-section, the
Trachylinae (or Cordyliotae), on the other hand, invariably bear true cordyli or auditory
clubs with endodermal otolites on the margin of the velum ; their tentacles are, moreover,
usually solid, tolerably stiff, and slightly contractile filaments, and their velum is thick and
compact. The Leptolinae mostly develop indirectly (by metagenesis), the Trachylinae
mostly directly (by hypogenesis). In both sub-sections there is an order with gastral geni-

REPORT ON THE DEEP-SEA MEDUSAE.

IX

talia (Anthomedusae, Narcomedusae), and an order with vascular genitalia (Leptomedusae,
Traehomedusae). Those differences are clearly shown by the following tables : —
§ 8. Survey of the two sections of Craspedotae.

Leptolin/E (Acordylite).

Craspedota? without auditory clubs or cordyli, with
very movable and extensible tentacles, which are
usually hollow.

Trachylin2E (Cordyliotse).

Craspedotae with auditory clubs or cordyli, with stiff
and slightly extensible tentacles, which are usually
solid.

Order I. Anthomedus.e, PI. I.

(System, p. 3, taf. i.-vii)

Genitalia gastral (reproductive organs originally in
the subumbral wall of the stomach), descent from
Tubularia polyps.

Order IL Leptomedosj;, PI. II.

(System, p. 3, taf. viii-xv.)

Genitalia vascular (reproductive organs originally in
the subumbral wall of the radial canals), descent
from Campanularia polyps.

Order III. Trachomedus^e, Pis. III. -VIII.
(System, p. 234, taf. xvi.-xviii.)
Genitalia vascular (reproductive organs originally in
the subumbral wall of the radial canals), usually
without peronia on the umbrella margin, never
with peronial canals.

Order IV. Narcomedus/E, Pis. IX -XIV.
(System, p. 299, taf. xix.-xx.)
Genitalia gastral (reproductive organs originally in
the subumbral wall of the stomach), peronia on the
umbrella margin, usually with developed peronial
canals.

§ 9. Orders of the Acraspedae (System, pp. 362, 449, 632). The section of the
Acraspedae or Scyphomedusse is divided into two sub-sections and four orders. The first,
older and lower sub-section, the Tesseroniae (or Tetraperiae), has a highly arched, usually
conical umbrella, and on its subumbral side four large perradial gastral pouches, separ-
ated by four interradial septa or cathamma, in whose subumbral wall the genitalia are
developed. The second, younger and higher sub-section, the Ephyroniae (or Octoperiae)
has, on the contraiy, a depressed, usually discoid umbrella, and on its subumbral side, a
very wide, flat gastral cavity, into which the four original perradial pouches have been
merged, and in whose subumbral wall the genitalia are therefore developed ; the four
interradial septa or cathamma have undergone retrograde formation, and are mostly lost.
A further distinction between the two sub-sections lies in the number of their character-
istic sense clubs or rhopalia. All Ephyronia or Discomedusse have eight or more sense
clubs (four perradial, four interradial, often several accessory), whilst the Tesseroniae have
only four sense clubs, or none at all. Of the three orders of the Tesseroniae, the Pero-
medusae have four interradial rhopalia, the Cubomedusae four perradial, and the Stauro-

(ZOOL. CHALL. EXP. — PART XII. — 1881.) M b

X THE VOYAGE OF H.M.S. CHALLENGER.

medusas no rhopalia (tentacles instead). These differences are shown by the following
tables : —

§ 10. Survey of the two sections of the Acraspedae.

Tesseronle (Tetraperias).

Acraspeda? without sense clubs, or with four sense
clubs (rhopalia). Umbrella usually highly arched,
conical ; four perradial gastral pouches large, sepa-
rated by developed cathammal septa.

Order I. Stadeomedusje, Pis. X V.-XVII.
(System, p. 363, taf. xxi.-xxii.)
No rhopalia ; simple tentacles instead.

Order II. Peromedus*:, Pis. XVIH-XXV.
(System, p. 396, taf. xxiii., xxiv.)
Four interradial rhopalia and four perradial tentacles
between (often eight adradial tentacles in addition).

Order DX Cubomedus^, PI. XXVI.
(System, p. 423, taf. xxv., xxvi.)
Four perradial rhopalia and four interradial tentacles
or bunches of tentacles between.

Ephyronle (Oetoperia;).

Acraspedae with eight or more (up to thirty-two) sense
clubs (rhopalia). Umbrella usually depressed,
discoid ; four perradial pouches small, usually
become part of the central stomach, by dissolution
of the cathammal septa.

Order IV. Discomedus;e, Pis. XXVII.-XXXH.
(System, p. 450, taf. xxvii.-xL)
Eight rhopalia or more, four perradial, and four
interradial (sometimes also several accessory).

Sub-Order I. Cannostomae.
GSsophagus simple, without free oral arms,
tacles usually short and solid.

Ten-

Sub-Order II. Semostomas.
Oesophagus cleft into four large perradial, folded oral
arms. Tentacles usually long and hollow.

Sub-Order III. Rhizostoma?.
(Esophagus represented by eight adradial dice-shaped
oral arms with numerous funnel openings. Central
mouth fused. No tentacles.

§ 11. Polyphyletic origin of the Craspedotae. The section Craspedotaa or Hydro-
medusas is probably a polyphyletic animal group ; several different groups of Craspedotas
having arisen independently of one another from several different groups of Hydropolyps.
This is corroborated by the fact that all Anthomedusae are descended from Tubularian
Polyps, all Leptomedusse from Campanularian Polyps. The phyletic divergence of these
two groups of polyps is probably very much older than the origin of the two correspond-
ing orders of Medusae. The Trachylinas or Cordyliotae seem to form a third independent
group of this section, the Trachomedusas and Narcomedusae. But these two orders may
also have originated independently of each other. The monophyletic origin of all forms
within the four orders of Craspedotae from a single common ancestral form, though not
certain, is still extremely probable (System, p. 359).

REPORT ON THE DEEP-SEA MEDUSAE.

XI

12. Hypothetical Ancestral Tree of the Craspedotse (Polyphylctic).

II. LEPTOMEDUS/E

III. TRACHOMEDUS.E

I. ANTHOMEDUS.E

(Ctenophor*)

Cladonemid.e
Dendronemidre

(SipnoNOPHOR^:)

Tiarid.e
Pandajidoe

Ctemria

Pteronemidse

Amphinemidfe
I

Protiaridse
(Protiara)

MARGELrD.E

Cladostylse

Haplostyls
(Cyteeis)

iEQUORIDa: GERYONIDiE

Polycannida? Carmarinida:

Zygocannidaa

Cannotid^;
Berenicidoe

Williadoe

Polorchidse

Oetocannidse
Octocanna

EucopiDa: Liriopidae
Polyotessae Liriantlm

Aglaurid^;
Persida;

Aglanthidas

IV. NARCOMEDUS.E

SOLMARIWE

Solmundinae
(Solmwndus)

Octotessae

Thaumantid^
Orchistoniida;

Melicertids

Laodicida?

Staurodisais

Eucopium

Tr.ACHYNEMIDiE

Pectyllida;

I
Marmanemidae

Aglantlia

Trachynema

Solmonetida;
(Solmoncta)

Solmissida?
(Solmissus)

Petasid^;
Olindiadae

I

I
Petachnida?

Codonidb

Tetranemal ancestral form:
Codonium*

Ancestral Polyps of the
Anthomeduspe :

TUBULARI.E

Thaumantid^;

Tetranemal ancestral form :
Tetrancma*

Ancestral Polyps of the

Leptomedusae:

Campantjlari^;

PETASIDi;

Tretranemal ancestral form :
Petasus*

JEaiNms
iEginodorida}

PEGANTHrD*

{Polycolpa)

iEginuridae

.ffiginetidaa
C&giiw.)

CUNANTHIDJE

Tetranemal ancestral form:
(Cunaiiiha*)

Ancestral Polyps of the Trachylina?

(Trachomedusa? and Narcomedusce) :

(Trachtlari^;?)

Hydropoltpi

Hydropolypi

-Scyphopolypi : Scyphosloma

ARCHYDRA (Ancestral form of all Cnidarise)

GastrjEA (Ancestral form of all Metazoa)

xu THE VOYAGE OF H.M.S. CHALLENGER.

§ 13. Hypothetical Ancestral Tree of the Acraspedse (Monophyletic).

IV.
DISCOMEDUSVE

II.

PEROMEDUSjE.
i * \

PeRIPHYLLID/B

IV. C. Ehizostom^i

CkambessidjE
Versuridj: Leptobrachida;

Stylorhizidse

Crossostomidaj

Pericolpidje
(Pericolpa)

Pericolpida?

{Pericolpa*)

[Ancestral form of

the Perornedusa?.]

Himantostomida!
Eucrambessida?

Pilemid^:
Eupilemidai

Stomolophidoa
Lychnorhizidae

III.

CUBOMEDUSiE

Chirodropid,e

TOREUMID2E

Polyrhizidje

Polyclonidae

Haplorhizidre

CHARYBDEID.fi

Tamoyidae

Archirhizida;
(Archirhiza*)

IV. B. Semostom^e

I.

STAUROMEDUSjE

ltjcernarid.e
Halicyathidse

Haliclystida?

Tesserid^
Depastridae

Tesseranthidae

(Tessera*)

[Ancestral form of

the Stauromedusa;]

ULMARID.E

Aurelida?

Sthenonidre

Ctaneid^;
Drymonemidae

Umbrosida?
( Ulinaris)

Flosoitlid^!
(Floseula)

Pateridas
I

Medoridoj
(Procyanca)

Pelagid^!
(Pelagia*)

IV. A. Cannostom^:

LlNERGIDffi

EpHYRiDiB Liuuchidae

Collaspidae

TTausithoidaa
I

Linanthida?
I

Palephyridae

(Ephyra*)

[Ancestral form of the

Discomedusa;]

Procharagmida;

(ProcJiaragma*)
[Ancestral form of
the Cuboniedusae]

Tesserid^:

Common ancestral form of all Acraspeda; or Seyphoniedusse
Tessera

rC o r a l l a :"
LAnthozoa.

-Scyphopolypi (Ancestral form of all Scyphopolypi : Scyphostoma)

Hydropolypi-

Hydropolypi

-ARCHYDKA (Ancestral form of all Cnidariaj)

Gastr.s:a (Ancestral form of all Metazoa)

REPORT ON THE DEEP-SEA MEDUSAE. X1U

§ 14. Monophyletic origin of the Acraspedse. Whilst the section of the Craspedotse
is probably of polyphyletic origin, and their chief groups have arisen independently of one
another, the conditions of relationship among the Acraspedse or Scyphomedusse is, on the
contrary, so close that the monophyletic origin of these sections may be admitted with the
greatest probability. The existing Tessera has even retained a very old primordial form,
which has the same significance for the section of the Acraspedse as Hydra has for the
whole tribe of the Cnidaria. The order of Stauromedusse in which the sense clubs or
rhopalia are completely wanting, is developed first of all from Tessera. The two orders
of Peromedusse and Cubomedusse, spring as two diverging principal branches from the
ancestral group of the Stauromedusse ; in the former the four interradial tentacles are
transformed into sense clubs or rhopalia ; in the latter, on the contrary, it is the four per-
radial tentacles. The fourth order of Acraspedse, the extensive group of the Discomedusse,
is distinguished by the fact that all the eight principal tentacles are transformed
into rhopalia ; it has probably originated from the Stauromedusse or Cubomedusse, per-
haps rather from the Peromedusse. The Cannostomse form the ancestral group among the
Discomedusse ; the Semostomse have probably arisen later from these, and the Rhizostomse
still later from the Semostomse (System, p. 450).

§ 15. Ontogenesis of the Medusse. The individual development or ontogenesis,
which, according to the biogenetic fundamental law, affords us the surest explanation of
the phylogenesis or the original development of the tribe, appears in the Medusse in
two distinct principal forms, as metagenesis and hypogenesis. Metagenesis (or indirect
development) includes the alternation of generations of the Medusse ; the Medusa is
formed here asexually by gemmation from a sessile polyp ; and the ova of the free-swim-
ming Medusse do not again develop into Medusa, but into the Polyp, which functions
as bud-bearing " nurse." On the other hand, hypogenesis (or direct development) of the
Medusa consists of the immediate development of the Medusse from the ova of the
Medusse ; the intermediate stage of the Polyp-nurse is wanting. As then the compara-
tive anatomy indubitably shows that the wandering, more highly developed Medusa
form is to be derived phylogenetically from the sessile Polyp form, we must consider
the metagenesis of the Medusse as their original or pabngenetic mode of development,
but their hypogenesis as the modified, shortened, and simplified cenogenetic form of
gemmation. It is the metagenesis, not the hypogenesis, which repeats in epitome the
method of historical development.

§ 16. Metagenesis or indirect development. The alternation of generations or meta-
genesis of Medusse must be considered their original form of generation or primary mode
of development, — their palingenesis; it appears in different forms in the two sections of
the class. The Craspedotse originate from Hydropolyps by lateral gemmation, the Acras-
pedse from Scyphopolyps by terminal gemmation. These two forms of alternation of

xiv THE VOYAGE OF H.M.S. CHALLENGER.

generations are not referable to one another, and have arisen independently of one another.
Among the Craspedotse, alternation of generation is the usual form of development of the
Leptolinse (or Acordyliaj), and the Polyp-nurses of the Anthomedusse are Tubularian Polyps,
whilst the Polyp-nurses of the Leptomedusse are Campanularian Polyps. On the other
hand, metagenesis only occurs rarely among the Trachylinse (orCordyliotse) (Lovenella clausa
is perhaps the Campanularian nurse of a Trachomedusa (?) — System, p. 653). Among the
Acraspedse, ontogenesis of the Tesseronise is still unknown. Alternation of generations
seems to be the usual form of development in the Ephyronice. The peculiar form of
terminal gemmation, by which the Discomedusa is developed from the Scyphopolyp nurse,
is, however, essentially different from the lateral gemmation by which the Craspedotse are
developed from the Hydropolyp nurse.

§ 17. Hypogenesis or direct development. Direct development without alternation of
generations, which in one word we call hypogenesis, must not be regarded as the original
form of generation in the Medusae, but as a secondary, shortened and simplified mode
of development, — as cenogenesis ; it has arisen by lapse of the alternation of generations.
Among the Craspedotse, nearly all Trachylinse (Trachoinedusse and Narcomedusse) develop
in this manner, but only a very few Leptolinse {e.g., the Cannotid Dipleurosoma =
Ametrangia, System, p. 637). Among the Acraspedse probably many Tesseronia (Stauro-
medusse, Peromedusse, and Cubomedusse) are similarly developed directly from the ovum ;
their ontogeny is, however, unknown as yet. Among the Ephyronia (Discomedusse)
hypogenesis is as yet only known to be constant in Pelagia (whilst the closely allied
Chrysaora undergoes metagenesis). Aurelia is usually developed with alternation of
generations, but in isolated cases without it, directly from the ovum.

§ 18. Medusse and Acalephse. Since the alternation of generations with Polpys has
been known in the Medusse, great difficulties have arisen in the classification of the Aca-
lephse ; and the natural class of Medusse, which can be so easily distinguished from other
classes of Cnidarise, has therefore sometimes been even abandoned by many more recent
authors. From the standpoint of the doctrine of evolution, however, these difficulties can
be easily solved, and the Medusse, at the same time reinstated as a class. According
to the present extent of our knowledge, it appears most logical, and at the same time,
natural to distinguish definitely the following five classes among the Acalephse : — Class I.
Polyps (Polypi) includes the common ancestral group of all Cnidarise ; it is divided first
of all into two sections, Hydropolyps and Scyphopolyps. To the Hydropolyps (without
tseniola) belong (a) the hypothetic ancestral form itself Archydra (also Gastrcea, closely
related to Hydra); (b) Hydropolyps without Medusa formation, and with simple genitaha
(Hydra, Clava) ; (c) Hydropolyps with spore-sacs or medusoid genitalia (Tubularise, Cam-
panularise) ; these spore-sacs or medusiform reproductive buds ("sporosacci")are Medusse,
which have undergone retrograde formation, without oral opening and without tentacles

REPORT ON THE DEEP-SEA MEDUSAE. xv

and organs of sense. To the Scyphopolyps belong : (a) the real ancestral form of the
Acraspedse (Scyphostoma) ; (b) the hypothetic ancestral forms of the corals (Procoral-
lium) ; (c) Scyphopolyps, which are probably propagated as such without Medusa formation
(Spongicola, Stephanoscyphus). Class II. Corals {Corolla, or Anthozoa) is phylogenetically
derivable from the Scyphopolyps (ProcoraUimri), probably a polyphyletic group (having
arisen at different times from several different groups of the latter). Class III. Medusae is
most probably polyphyletic ; the section of the Craspedotae has likely arisen from several
groups of Hydropolyps (Polyphyletic, § 11), but the section of the Acraspedse from a
single group of Scyphopolyps (Monophyletic, § 14). Class IV. Ctenophora is probably
monophyletic, having proceeded from a group of Anthomedusae (Cladonemidae) (Ctenaria,
System, p. 107). Class V. Siphonophora is probably polyphyletic, having proceeded from
several forms of Anthomedusas (Codonidse, Sarsiadae, System, pp. 14, 20, &c). The
Siphonophora are polymorphic Medusae cormi or colonies, whose associated personae have
become differentiated by division of labour, and assumed very different forms.

§ 19. Ectocarp and endocarp Medusas. As the Craspedotse are more widely separated
by the foregoing distinctions (above all by the absence of the gastral filaments and by
the ectodermal genitalia), it is proposed to dissolve the class of Medusae entirely, and to
divide the whole tribe of the Acalephse or Cnidarise into two principal groups, of which
one (Ectocarpse) includes the groups just mentioned, the other (Endocarpse) the Acras-
pedse, the Scyphopolyps, and the Corals (with gastral filaments and with endodermal
genitalia). This proposition seems entirely justified from a phylogenetic point of view,
and we would accept it unconditionally, if we were in a position to carry out a
phylogenetic system of the Cnidarise completely and with certainty. Unfortunately, this
is not the case at present. The most probable admission at present is, that the ancestral
group of the Acalephse (the primitive polyps, Archydras, § 18) were early split up into
the two diverging tribes of the ectocaqD Hydropolyps (without taaniola) and the endocarp
Scyphopolyps (with taeniola). The Craspedotse (with the later side branches the
Ctenophora and Siphonophora) issued from the former, the Acraspedae and Anthozoa
(Corals) from the latter. Only, as, moreover, a polyphyletic origin has now become
more probable for the majority of the said classes of Cnidarise, it seems more accurate
for the sharp definition of these classes and their logical arrangement, to abandon at pre-
sent carrying out a phylogenetic system and to define the said five classes to the extent
known : — (1) Polypi, (2) Medusse, (3) Siphonophora, (4) Ctenophora, (5) Coralla, If,
on the other hand, we prefer to take their phylogenetic conditions as the fundamental
plan of their systematic classification, it would be carried out according to the following-
table : —

xvi THE VOYAGE OF H.M.S. CHALLENGER.

§ 20. Survey of the two ancestral branches of the Acalephse.

I. First Ancestral branch of the Acalephse

(without gastral tseniola, with ectodermal genitalia).

Acaleplue ectocarpse (Intoeniolae).

II. Second Ancestral branch of the Acalephse

(with gastral tamiola and endodermal genitalia).

Acalephce endocarpse (Tseniolata?).

1. Hydropolyps (common ancestral group of all Aca-

lephse, and first of all of the Intseniolae).

2. Craspedota? (Hydroniedusa?, which have origin-

ated from sessile Hydropolyps by adaptation
to a free-swimming mode of life).

3. Ctenophorse (an early side branch of the Craspe-

dots — Anthomedusae).

4. Siphonophorre (swimming colonies of Craspedotas —

Anthomedusas), with polymorphism of the per-
sonse.

5. Scyphopolyps (ancestral group of the tamiolataa,

derived from a branch of the Hydropolyps).

6. Acraspeda? (Scyphomedusaj, originated from sessile

Scyphopolyps by adaptation to a swimming
mode of life).

7. Coralla (Anthozoa, the principal group of the

sessile Tseniolata, probably sprung from several
branches of the Scyphopolyps).

§ 21. Individuality of the Medusa. The tectological value, or the individual stage
of form of the fully developed and mature Medusa is in every case that of an
inarticulate transversely axonial (or radiate) persona (" persona inarticulata staur-
axonia "). Each Medusa has, therefore, a vertical principal axis (with oral and aboral
pole) and two or more horizontal transverse axes, perpendicular to it, with twice as
many radii. The number of these radii (usually four) corresponds to the number of
the radial sections or parameres which compose the body of the inarticulate persona and
which only touch by their axial edges in the vertical principal axis of the body. Those
Medusae which form colonies or cormi, undergo at the same time considerable organo-
logical differentiations in consequence of extensive polymorphism of their personse and
become separated from the true Medusse as a special class : the Siphonophora.

§ 22. Fundamental form of the Medusa. In all Acraspedse and in the majority of
Craspedotae, the regular pyramid is the geometrical basis-form of the mature body, and
the principal axis of the Medusa body (which is vertical in the normal position of the
animal) is the axis of the pyramid ; its upper (aboral) pole corresponding to the point,
whilst its lower (oral) pole, with the oral opening, falls in the middle of the base of
the pyramid. The angles of the regular pyramid (at least four) are transected by
the primary transverse axes (with twice as many radii of the first order), whilst the
middle lines of the lateral surfaces of the pyramid correspond to the indifferent radii
of the second order.

REPORT ON THE DEEP-SEA MEDUSAE. XVll

§ 23. Primary cardinal number of the Medusa. In all Acraspedae, and in the majority
of Craspedotse, the original cardinal number of the Cnidariae (4) is invariably retained
(apart from numerous individual exceptions). The geometrical fundamental form of the
ordinary quadripartite Medusa is the quadrate pyramid or " quadrangular regular pyra-
mid " ; its base forms a square. Only two primary transverse axes exist (the two
diagonals of the square), and these cross at right angles.

§ 24. Individual digressions from the primary or typical cardinal number. It is not
unusual to find individual abnormalities of the fundamental number (more frequently in
many groups or many species, more rarely in others), so that instead of four radii there
are six or eight, less commonly other numbers (5, 7, 9, &c.) ; in single species (e.g.,
Aurelia aurita) this inclination to individual variation of the fundamental number is very
great ; in some groups (e.g., Cannotidae, System, taf. ix.) it is constant in many species
or genera, so that here of closely allied species or genera, some are quadripartite and
others sexpartite. In some other groups (iEquoridae, Cunanthidae, Peganthidae, Solmaridae)
the fundamental number is very large and indefinite ; it may mount up to above a
hundred (e.g., JEquorea forskalea. jE. ciliata, Mesonema ccsrulescens, M. dubium, &c).
The fundamental number is more inconstant the higher it rises, and therefore more
unequal in the different individuals of a species.

§ 25. Secondary fundamental numbers of Medusae. Next to the primary fundamental
number 4, in the Medusae only 6, 8, and 12 are prominently significant as secondary
fundamental n ambers. They become hereditary in many species and groups of species,
and thereby acquire a systematic significance. On the other hand, the uneven numbers,
5 and 7, do not appear constantly in any single species of Medusa, but only as individual
variations, and so do 9, 10, and 11. As soon as we get beyond the number 12 the
fundamental number especially loses all morphological and systematic significance, as it
then becomes inconstant and variable, and more so the higher it rises (iEquoridae,
Cunanthidae, Peganthidae, Solmaridae). Moreover, as the fundamental number 8, which
often recurs, and is constant in many groups, has arisen by duplication of 4, as the more
unusual number 12 from duplication of 6, the most important secondary fundamental
number next to the primary 4 really consists of 6 ; instead of the normal 2 transverse
axes of the regular pyramid 3 are formed abnormally, and the hexagonal regular pyramid,
therefore, appears instead of the quadrangular.

§ 26. Central and peripheric numbers. In all Medusae it is only the central part of
the body, pre-eminently the stomach, never the peripheric parts, which are the criterion
for determining the true fundamental number or homotypical number. The peripheric
parts, especially the umbrella margin with its appendages, show higher numbers than
the central part in the majority of Medusae. In most cases these higher numbers of the
peripheric parts have arisen either by multiplication from the lower fundamental number
of the central part or by regular multiplication according to definite conditions of pro-

(ZOOL. CHALL. EXP. PART XII. — 1881.) M C

XVin THE VOYAGE OF H.M.S. CHALLENGER.

gression. On the other hand, it rarely happens that the central part is four, and the
peripheric part of the umbrella six {e.g., Polyclonia frondosa). That the central umbrella
usually retains the fundamental number by inheritance, whilst the peripheric part
varies in several ways, is explained by the fact that the latter is subject in a higher
degree to adaptation. (The fundamental number of all Rhizostomae, for example,
remains 4, with four oral pillars, four genitalia, &c, although they have all eight
arms and from eight to twelve sense clubs, as well as a very variable number of marginal
lobes.)

§ 27. Radii of the first to the fourth order. The radial structure of the Medusae (like
that of most radiata) is caused by the division of the growth of the central body (originally
uniaxial in the gastrula) into different meridian planes. As the growth is more energetic
in definite meridian planes, or radial planes (which touch in the common central prin-
cipal axis), and leads to the development of new organs, the interlying radial planes con-
tinue indifferent or opposed, and usually in the middle between these energetic rays of
growth. In this way an antithesis arises first of all between radii of Order I. and Order
II., which we shall designate shortly " perradii " and " interradii." Special organs very
often arise in the middle between the perradii and the interradii, and these then lie in the
radii of Order III., the "adradii." Finally, we can distinguish in many cases radii of
Order IV., or subradii, which lie in the middle between the eight adradii and the eight
principal radii. In the "principal radii" we include the four perradii and the four
interradii, whilst, in contrast to these, we term all other possible radii " succursal or
secondary radii." Our distinction of these four orders of radii is of great importance, not
only for the architecture of the Medusae, but for the promorphology of most other
" radiata" ; it allows us to designate, in a single word, the most important conditions of
position and relation of the organs with mathematical sharpness and precision. For
example, Ephyra, the important ancestral form of all Disconiedusaa (fig. A), has four
perradial oral lobes and limbs of the oral cross (as), four interradial genitalia (s) and
filaments or phacelb (/), eight adradial tentacles (ta), and sixteen subradial marginal
lobes (/).

§ 28. Parameres and antimeres. As in all Acraspedse, and in the majority of the
Craspedotse, 4 is the normal typical fundamental number of the body, the latter conse-
quently consists of four parameres or " radiate parts," which touch in the common vertical
principal axis of the body. These four parameres are originally congruent, so that a prin-
cipal organ comes on each paramere, — an oral lobe, a quadrant of the stomach, a radial
canal or a radial pouch, a septum or cathamma between the pouches, a primary tentacle, and
so forth. Each paramere has a cbpleuric fundamental form (or a strictly " bilaterally
symmetrical " form), and therefore again consists of two equally symmetrical halves, the
"counterparts" or antimeres. These two antimeres or counterparts comport themselves
the same as the symmetrical halves of the body of all higher (dipleuric) animals. We

REPORT ON THE DEEP-SEA MEDUSAE.

XIX

can also distinguish three axes or " euthyna " in each paramere ; the longitudinal
axis of the paramere coincides with the central principal axis of the whole body ; the
sagittal axis (or " dorsoventral " axis) of the paramere is the perradius ; its lateral axis
(or transversal axis) lies tangentially, and touches the two adjacent parameres with its two
poles (" right or left pole ").

Perrad.

(I.O.J

&._— .... :
„

^^v ' x i$

^^^^^fc^ T^^^'m

>:::::::■

^~ .^ r-^"^^ssS

/. .-^.

— ■■ ^ ^

-,79? * - _ sT?T2-z._r*, f

«,>• J=

r^j^":c^ggm

2 ^^

O1 - - - - - -wk B-B--M

c i ,Adra.d.
hubrad. im0)

J fh O 1 ,Jntermd,
7 oubrad. Au.o.i

Adrad.

' (III.O.I

PerraM
(io.)

Fig. A. Zoncphyra pclagica (Diseomedusa, Ephyrida).

Suburabral view, giving the four orders of transverse axes (with twice as many radii). The oral cross
(as) and the four perradial sense clubs (o1) lie in the four perradii (Order I.). The gastral filaments
(/), the genitalia (s), and the four interradial sense clubs (o°) lie in the four interradii (Order II.).
The eight tentacles (to) and the tentacular coronal pouches (bt) lie in the eight adradii (Order III. ).
The sixteen marginal lobes (I) lie in the sixteen subradii (Order IV.). Sixteen bifurcate coronal
pouches, eight tentacular (6/) and eight rhopalar (ho) radiate from the central stomach (g).

§ 29. Radial planes or meridian planes. In the Medusa?, as in all regular or symmetri-
cally radially constructed animals, we can distinguish a number of radial or meridian
planes, which have a high tectological and promorphological significance, and which are
defined by the position of the vertical principal axis and of one of the horizontal trans-
verse axes. As in all Acraspedas, and in the majority of Craspedotse, the body has the
fundamental form of the quadrate pyramid, we can distinguish in them four principal

XX THE VOYAGE OF H.M.S CHALLENGER.

meridian planes, which are defined by the position of the four perradii and the four inter-
radii, and which stand in definite relation to the four parameres or pairs of antimeres
(comp. woodcut, fig. A). The four perradii lie in the two primary meridian planes (or
the " radial planes of Order I.") whilst the four interradii lie in the two secondary
mcridan planes (or the " radial planes of Order II."). Each of the two primary meridian
planes is therefore the middle plane (or saggital plane) of each two opposite parameres and
at the same time the boundary planes between the two symmetrical antimeres of each
paramere. On the other hand, each of the two secondary meridian planes is the boundary
plane (or septal plane) of the two adjacent parameres, and at the same time the boundary
plane of their two contiguous antimeres. If the four corners of the quadrate pyramid corre-
spond to the four perradii, its diagonal planes (in which each two opposite corners lie) are
the primary meridian planes, whilst the secondary meridian planes are formed by the
middle lines of each two opposite lateral surfaces of the pyramid.

§ 30. Regular and amphitect pyramids. Whilst all Acraspedae and the majority of
Craspedotae retain the original primary fundamental form of the regular pyramid, in a
large number of Craspedotae (but in no Acraspedae) it passes into the secondary funda-
mental form of the bisected or amphitect pyramid. In the former, all the four parameres
of the body are completely congruent, whilst in the latter they are only congruent in
pairs, as each two adjacent parameres are symmetrically similar. Whilst the base of
the regularly quadrangular pyramid is a quadrate, that of the amphitect quadrangular
pyramid represents a rhombus. As in the amphitect pyramid the two primary transverse
axes are unequal in size, so, of the four parameres, each two adjacent are symmetrically
similar, each two opposite are congruent. In these amphitect Craspedotae it is usually
the unequal development of the tentacles which first causes the alteration of the regular
fundamental form. Of the four primary tentacles, the two opposite are much larger or the
only ones developed, whilst the other two, alternating with them, are smaller or quite
rudimentary (e.g., Thamnostylus, PL I. ; Discodonium, System, taf. i. fig. 6 ; Ctenaria,
System, taf. vii. figs. 3, 5 ; Dissonema, System, taf. viii. fig. 3 ; Dipetams, System,
taf. xviii. fig. 2 ; JEginella, System, taf. xx. fig. 16). The promorphological con-
ditions of the rhomboid pyramid are entirely the same in these amphitect four
Craspedotae as in the Ctenoplwra. In a few rare cases, the hexagonal amphitect pyramid
also appears along with the quadrangular, the same promorphism which distinguishes
most corals (e.g., Dipleurosoma, System, taf. ix. fig. 9).

§31. Dipleuric or zeugite pyramids. The "pair pyramid," or zeugite pyramid,
appears much more rarely than the amphitect form, along with the predominating
regular pyramid as the geometrical fundamental form of the Medusae. In such cases the
whole Medusa persona has the same promorphological conditions as each one of the four
parameres in the ordinary regular Medusa persona. This condition is only found strongly
expressed in the Anthomedusa family of the Codonidae, and characterises there the

REPORT ON THE DEEP-SEA MEDUSA.

XXI

special subfamily of the Euphysidse. In it, of the four perradial tentacles, three are
rudimentary and transformed into marginal ocelli ; the fourth tentacle only is developed,
and therefore so much the more strongly. It is simple in Euphysa and Steenstrv/pia
(System, taf. ii. figs. 8-14), but split into two to three filaments in Amphicodon (System,
taf. i. figs. 7-9) ; whilst in the former the fundamental form of the umbrella is only altered
a little, in the latter (as also in Ilybocodon) it is considerably transformed symmetrically,

Fig. B. Cannorhiza connem (Discomedusffi, Versurids;).

Subumbral view of the umbrella. The arm disk with the eight oral arms is removed as the four per-
radial arm pillars (ab), which connect the umbrella disk and the arm disk, are cut through, (oi)
Inter sense clubs, (nm) Umbrella margin (turned over inwards), (s) Genitalia, (ira) Gelatinous
cross of the gastrogenital membrane, (gg), (gh) Peripheric limbs of the gelatinous cross, (ug)
Peripheric umbrella corona, (cc) Coronal canal, (cd) Pillar canals, (ca) Adradial canals, (ci)
Interradial canals, (ep) Perradial canals.

and distinctly dipleuric or zeugite. The whole umbrella is here bilaterally compressed
and divided by a sagittal plane (in which the curved gastral axis lies) into two sym-
metrical equal halves, a right and a left. The lateral transverse axis is equipolar and
shorter than the polar sagittal transverse axis, which has often a tentacle at the one
pole, and often has Medusae buds at its base. The further the tentacle and its group
of buds extends up the umbrella at this point, the more distinct the zeugite form becomes.

xxii THE VOYAGE OF H.M.S. CHALLENGED

This fundamental form has become predominant in the Medusoid personae of the
Siphonophora colonies. It appears also in the parasitic Mnestra, and is sometimes
faintly indicated in some Cubomedusae (Charybdea).

§ 32. Perradii, or transverse axes of Order I. In the ordinary regularly quadripartite
Medusa?, the four perradii or " upper radii " lie in the two transverse axes of the
quadrate pyramid perpendicular to each other, in the sagittal middle hne of the four
parameres, between these two antimeres. In most Craspedotae the four oral angles, or
oral lobes, the four radial canals, the four primary tentacles, and the four genitalia lie in
the four perradii (or with dichotomised genitalia in the middle line of the perradii between
the halves). In all Acraspedse the four limbs of the oral cross and of the gastral cross
(fig. B, ivx) correspond to the four perradii, also the four oral lobes and the four oral
pillars or primary oral arms (ab), and also the middle line of the four lobed primary
radial pouches : in the Staurornedusae and Peromedusae in addition to these the four
tentacles Lie in the four perradii, and in the Cubomedusae and Discomedusae the four
primary sense clubs (fig. B, op). In all Medusae the perradii are originally the zones of
the most active life, of the strongest growth, and most complicated differentiation, with
preponderating tendency to centrifugal development of the organs.

§ 33. Interradii, or transverse axes of Order II. In the quadripartite Medusae the
four interradii lie exactly in the middle between the four perradii, and, therefore, intersect
the latter at an angle of 45 degrees; they therefore lie at the same time in the boundary hne,
between the four parameres. In the quadrate Craspedotae the four interradii form the
geometrical middle line of the four injected oral archings (between every two perradial
oral lobes), and also of the four broad cathammal plates (between every two radial canals) ;
in the octonemal Craspedotae (with eight tentacles) the four primary tentacles He in the
perradii, the four secondary in the interradii. In all Acraspedae the four injected oral
angles and oral columns lie in the four perradii, also the four fundamental taeniola or
gastral ridges, and the four primary gastral filaments (or groups of filaments), and, finally,
the four important cathamma (the four primary septal nodes or septal ridges). The
interradii, moreover, form the middle line of the four genitalia in most Acraspedae (fig.
B, s), and also of the four pair of genitalia in such Acraspedae in which the latter are
divided into two halves. In all Medusae the interradii are next to the perradii, the
zones of the most intense growth and of the most important differentiation, but the
tendency to development is predominantly centripetal with them, whilst with the
perradii it is centrifugal.

§ 34. Adradii, or transverse axes of Order III. The eight adradii lie in the middle
between the four perradii and the four interradii of the quadripartite Medusa ; they
halve the angles between the former and the latter, and intersect the two at an angle
of 22|> degrees. They consequently lie at the same time in the median planes of the
eight antimeres. If, on the one hand, we designate the four radii of Orders I. and II.

REPORT ON THE DEEP-SEA MEDUSAE. xxiii

the principal radii, on account of their prominent morphological significance, radii of
Orders III. and IV. belong, on the other hand, to the succursal or secondary radii, which
can only claim a subordinate value as compared with the others. In most Craspedotae the
eight adradii are of no special value; they are often without organs, or only bear the eight
tentacles of Order III. They are distinguished only in the vesiculated Leptomedusae
(Eucopidse and iEquoridas), as the eight typical " velar marginal vesicles " of this group
lie in them (fig. C, ov, System, taf. xi., xiii.). The eight adradii are of much greater
importance in the Acraspedaa. In these the eight hollow marginal " arms " of the
Lucernaridae and the homologous eight oral lobes of the Pericolpidae lie in the eight
adradii, also the eight marginal pouches of the Charybdeidas and the eight tentacles of
the Ephyra.

§ 35. Subradii, or transverse radii of Order IV. The sixteen subradii lie in all quadri-
partite Medusae in the middle between the eight adradii on the one side and the eight
principal radii (four perradii and four interradii) on the other; thirty -two angles of ll£
degrees remain between the former and the latter. Contrasted with the positive signi-
ficance of the principal radii, which in all Medusas mark the active meridian planes of
development, and the regular positions of the most important organs, the subradii have,
at most, a completely negative value ; they mark those meridian planes of the body which
of all comport themselves the most passively and indifferently. In most Craspedotae no
special organs lie in them, excepting in some Narcomedusae, as, for example, in AUginura
(Pis. XIII., XIV.), where the sixteen internemal pouches (with the genitalia) and at the same
time the sixteen auditory clubs of the umbrella margin lie subradially. In the majority
of the Acraspedae, the sixteen subradial planes are distinguished by the absence of ab1
organs. In a few groups of this section only they mark the site of isolated marginal
organs. For example, the sixteen marginal lobes of the Periphyllidae, among the Pero-
medusae (Pis. XVIII. , XIX.), and of the Ephyridae, among the Discomedusae (woodcut,
fig. A, I) lie subradially. The latter are so far of importance that the marginal lobes
of all Discomedusae have arisen from them : indeed, it is advantageous for the more
accurate morphology of this division to distinguish two groups among the sixteen
subradial " Ephyra lobes " ; the eight Ephyra lobes, enclosed in pairs by the perradii, are
corradial, whilst those enclosed in pairs by the interradii are exradial.

§ 36. Umbrella disk and umbrella corona (" discus umbrellae" and " corona umbrellae ").
In all Medusae the most important groups of organs of the body are divided in such a ■
way that a certain regularity seems common to them all. We can recognise especially
in distinct contrast two principal parts of the body between the central umbrella disk
(" discus umbralis ") and the peripheric umbrella corona ; the former contains the larger
and most important part of the vegetative organs, the latter, on the contrary, the prepon-
derating and most important part of the animal organs. The stomach and mouth, as well
as the many important organs developed round the mouth (oral lobes, or d arms, &c),

XXIV THE VOYAGE OF H.M.S. CHALLENGER.

belong to the central umbrella discus. The peripheric umbrella corona, on the other
hand, is characterised by the broad coronal muscle of the subumbrella, and, above all, by
the umbrella margin, on which the central nerve system, and also the sense organs and
tentacles lie, besides these the velum in the Craspedotse and the lobe corona in the
Acraspedse.

§ 36. Topographical antitheses. It is indispensable for the clear, and detaded ana-
tomical description of the Medusae, to avoid the usual, but indefinite and dubious terms,
" inner and outer, upper and lower " parts, &c, and instead of those to use definite topo-
graphical terms. For this purpose we represent the Medusa in its usual natural position
with the vertical principal axis, the convex umbrella surface turning upwards, the concave
umbrella surface turned downwards ; the umbrella margin forms the boundary between
them, in the more bmited sense, the free velum margin in the Craspedotas, the margin
of lobes in the Acraspedaa. All the parts lying above this free margin (or the upper con-
vex surface) we term dorsal or exumbral, all lying below it (on the lower, concave surface),
ventral or subumbral ; the marginal organs lie between the two on the umbrella margin.
With regard to the two poles of the vertical principal axis, we term all the central parts
which are turned towards its upper pole or apical pole " aboral," and those which are
directed towards the lower pole or oral pole " oral." Finally, as regards the two poles of
the radii or transverse axes, we name all parts of it which approach the central principal
axis " proximal parts," whilst those which turn towards the peripheric margin are " distal "
parts.

§ 38. Organic systems. All the different organs which are developed in the Medusae
may be divided into two large organic systems, the neurodermal and the gastrovascular
system. The neurodermal system includes preferably the animal organs and apparatus ;
the umbrella with its exumbral umbrella-covering and subumbral muscular plate, the
umbrella margin, with the most important organs of animal life, the central nervous
system, the tentacles and the organs of sense. The gastrovascular system, on the other
hand, consists principally of the vegetative organs of nutrition and reproduction, of the
central principal intestine (with stomach and mouth), and of the peripheric coronal
intestine (with pouches and canals), and also of the genitalia or reproductive glands,
which are invariably developed in the subumbral wall of the gastrovascular system (some-
times from the ectoderm, sometimes from the endoderm). Taken altogether, the neuro-
dermal system finds its most important site of formation in the peripheric umbrella corona,
the gastrovascular system in the central umbrella disk ; the former is preferably
ectoblastic, the latter endoblastic.

REPOET ON THE DEEP-SEA MEDUSAE. xxv

II. GENERAL HISTOLOGY OF THE MEDUSAE.

§ 39. Primitive germinal layer (" blastoderma "). In all Medusae as in all other
Metazoa, the aggregate cells of the developed body are descendants of the homogeneous,
indifferent " segmentation cells," which are formed by repeated division of the fecundated
egg cells, and which compose, first of all, the solid multicellular "mulberry germ"
("morula"). As fluid gathers inside this solid spheroidal accumulation of cells, and its
homogeneous cells appear on the upper surface, this important hollow sphere, the
"germinal vesicle" or "vesicular germ" ("blastosphera" or "blastula"), is originated,
whose wall is composed of a single, simple layer of cells. This simple cellular membrane
itself is the germinal membrane (" blastoderma") or the " primitive germinal layer." As
the hollow sphere then forms a depression at one point of its upper surface, and this
depression always deepens, the germinal membrane becomes invaginated, and thus
differentiates into the two primary germinal layers composing the " gastrula." As the
formation of the gastrula by invagination of the blastula in the Medusas has been observed
in very different groups, we may assume that it happens universally in this class, and
supposed exceptions (e.g., Geryonia) are founded on erroneous observations.

§ 40. Primary germinal layers ("ectoderma" and " endoderma"). The two primary
germinal layers, which first of all arise from the primitive germinal layer, have the same
fundamental morphological significance for the Medusas as for all other Metazoa (gastrasa
theory). As these two layers regularly recur now in the gastrula of all Metazoa, we may
assume that they have been transmitted by inheritance to all the groups from their
common ancestral form, the Gastrcea. According to the fundamental biogenetic law, they
therefore appear to be constant in the gastrula of all Medusas, which first develop by
invagination of the blastula. The inner or vegetative germinal layer, the intestinal layer
("endoderma," or " endoblastus"), limits the cavity of the primitive intestine, as a
simple nutritive cell layer, whilst the outer or animal germinal layer, the dermal layer
(" ectoderma," or " ectoblastus "), covers and protects the former layer from the outside,
as simple sensitive cell layer. In Medusas generally, the cells of the two primary
germinal layers (both the inner and the outer) are flagellate, high, cylindrical cells, each
of which bears a single, long vibrating flagellum. Whilst the vibrating flagella are
constant on the epithelial surface in the majority of the endoderm cells, they are lost in
the majority of the ectoderm cells. The two great organic systems of the Medusas stand
in definite relation to the two primary germinal layers, just as they are distributed over
the two principal sections of the body, the central umbrella disc, and the peripheric
umbrella corona. The more numerous and most important parts of the neurodermal
system arise from the ectoderm, those of the gastrovascular system, on the contrary,
more usually from the endoderm. During the development of the Medusas from the
gastrula a histological differentiation of the two primary germinal layers appears every-

(ZOOL. CHALL. EXP. PART XII. 1881.) M d

xxvi THE VOYAGE OF H.M.S. CHALLENGER.

where, which leads, as in higher animals, to the formation of different tissues, and these
tissues are regularly arranged in four strata, which in a certain sense may be considered
as " secondary germinal layers."

§ 41. Secondary germinal layers. If we consider the histological differentiation and
the structure of layers connected with it, of the organism of the Medusae as a whole, and
if we disregard the close connection of the different " principal layers " and " secondary
layers," we can universally distinguish accurately four layers, which follow one another
from the exterior, inwards : — (1) the dermal plate or dermal covering (" lamina chrotalis,"
" chrotoderma"), the layer of cells which covers the whole outer surface of the body in
continuous connection and passes into the endoderm at the oral margin ; (2) the muscular
plate (" lamina muscularis," " myoderma "), the thinner or thicker muscular layer, which
proceeds from the ectoderm and is chiefly spread over the concave lower side of the
umbrella ; (3) the connective plate (" lamina connectiva," " colloderma "), the gelatinous
or cartilaginous mass secreted from the endoderm, which as a thicker " gelatinous
umbrella " forms, according to volume and weight, the principal mass of the body in all
Medusae, but which is also found as the thinner " supporting lamella " in the subumbrella
and the tentacles ; (4) the intestinal plate or intestinal epithelium (" lamina gastralis,"
" gastroderma "), which lines the whole inner upper surface of the umbrella in continuous
connection, and passes into the ectoderm at the oral margin. Although the muscular
plates are for the most part produced from the ectoderm, and the connective plates, on
the other hand, from the endoderm, we must distinctly remember that in isolated cases it
is reversed, and muscles are formed from the inner germinal layer and supporting plates
from the outer.

§ 42. Two opposite views may be held in histologically judging the organism
of the Medusae. On the one hand, the Medusae may be considered as diblastic animals,
as, in the majority of them, all the tissues which appear between the two primary ger-
minal layers remain in close connection with them, are referable with certainty to one of
the two, and only acquire a slight amount of independence. But, on the other hand, we
may consider part of the Medusae (and in a certain sense all of them) as mesodermal (tri-
blastic or tetroblastic animals), as in certain parts of the body (and in some Medusae to a
great extent) independent tissues are really secreted between the outer and the inner ger-
minal layer, and so form a middle germinal layer (" mesoderma "). The following tissues
may be pre-eminently regarded as independent mesodermal tissues : — (l) the gelatinous
tissue of the umbrella as soon as it contains independent cells ; (2) the chordal tissue in
the axis of the solid tentacles ; (3) the muscular tissue of isolated, especially strongly
developed muscles ; (4) the nervous tissue in a part of the nerve centres and the organs
of sense. In a secondary degree, but less accurately, the following may claim to be
termed mesodermal tissues : — (1) the reproductive tissue ; (2) part of the urticating
tissue (the subepithelial urticating organs) ; (3) all subepithelial muscles ; (4) all

REPORT ON THE DEEP-SEA MEDUSAE.

xxvu

subepithelial nervous plexus. The latter, however, want the complete histological
independence and the entire separation from the mother-epithelia, already attained by
the former. On the whole we find autonomic mesodermal formations chiefly in the
higher and larger Acraspedae, in which both the volume of the body and the organological
separation have reached a very high grade, whilst they remain at a much lower stage in
the smaller Craspedotaa, which are much lower in this respect. If then isolated organs
are found on definite parts of the body, in which the different forms of tissue of the
animal's body have attained the same high and independent formation, as in the higher
animals, there is nothing to prevent us terming these secreted layers of tissue true
" secondary germinal layers" (even though these are only developed locally). The two
middle plates, the ectodermal muscular plate, and the endodermal connective plate, may
be classed together as mesoderm according to the following diagram.

Diblastic Theory.

Tetroblastic Theory.

Triblastic Theory.

Primitive germinal
layer, " Blasto- ■
derma."

I. Primary dermal layer,
" Ectoderma," s.a. ■
("Ectoblastus").

II. Primary intestinal
layer, "Endoderma," -
s.a. ("Endoblastus").

1. Secondary dermal plate

("Chrotoderma"). }

2. Muscular plate ("Myo- (

derma ").

3. Connective plate ("Col-
loderma ").

4. Secondary intestinal ]
plate (" Gastro V

derma "). j

External germinal layer,
" Ectoderma," s.st.

Middle germinal layer,
" Mesoderma."

Inner germinal layer,
" Endoderma," s.st.

§ 43. Differentiation and teleosis of the tissues. The great and general interest pre-
sented by the histological structure of the body of the Medusae does not only lie in the fact
that we can distinguish the origin of the four secondary germinal layers from the two
primary, and especially the derivation of the mesoderm from the two primary germinal
layers, more certainly and clearly in them than in the higher Metozoa, but also that in them
we can more clearly recognise the mechanical causes of these fundamental processes.
These mechanical causes, on the one hand, are the physiological division of labour of the
cells and the differentiation of the tissues proceeding from it, and, on the other hand, the
physiological perfection of the cells and the progressive development or teleosis of the
tissues resulting from it. If these processes of development continue to be carried on now
by inheritance in the ontogenesis of the Medusae, the result has been originally brought
about in their phylogenesis according to the laws of the theory of selection.

§ 44. Primary and secondary tissues. The importance of the Medusas for general his-
tology lies chiefly in this, that within this class a long series of histological differentiations

xxviii THE VOYAGE OF H.M.S. CHALLENGER.

and teleoses are developed from the simplest beginnings, step by step before our
eyes. Whilst at definite parts of the body (that is in many higher and larger Medusas) all
the four principal forms of the animal tissues are already secreted as independent layers,
they appear in other parts of the body (that is, in many smaller and lower Medusa?) still
in a dependent form, as mere appendages of a single fundamental tissue, the epithelium.
The most different degrees of formation of tissue are represented beside each other in
genetic connection, within this long series of perfection and differentiation, so that the
most important forms of the higher tissues are to be found here " in statu nascenti." In
this respect the Medusas furnish an excellent argument in favour of the tenet, recently
brought forward in the gastrasa theory, that there is only one primary tissue, the
epithelial tissue, and that all other forms of tissue have arisen secondarily from it.
The simplest and phylogenetically oldest form of this primary tissue is the blastoderm
of the "blastula," this simple single-layered epithelium, which alone forms the wall of
this hollow sphere in the germ of all Medusas, in the same way as it does in the germ
of all other groups of Metazoa. When the two-layered gastrula is formed by in-
vagination of the blastula, the blastoderm (or the simple " primitive germinal layer ")
of the former is divided into the " primary germinal layers " of the latter, which
are likewise simple epithelia. All other formations of tissue (connective, muscular,
and nervous tissues) have arisen, both ontogenetically and phylogenetically, from
these two epithelia.

§ 45. Epithelia] tissue (" tela epithelialis "). The protective tissue or epithelium, which
in the gastrula of the Medusas, as of all Metazoa, is formed first of all by the simple
tissue of the multicellular germ, in the mature and developed Medusa, covers firstly, as
outer covering (" ectoderm "), the whole upper surface of the body; and secondly, as inner
covering (" endoderm "), the whole inner surface of the gastrovascular system. These
coverings are everywhere separated from one another by secondary formations of tissues,
secreted between them, and only pass uninterruptedly into one another at the oral mar-
gin. This oral margin (am) is identical with the " primitive oral margin " of the gastrula
or the '" invagination opening" of the invaginated blastula. The inner covering ("epi-
thelium endodermale ") shows far simpler and more uniform conditions of formation in
both coverings. However, most of the differentiations recur in it, which appear more
expressed and varied in the outer covering (" epithelium ectodermale "), corresponding to
its manifold adaptations and relations to the outer world.

§ 46. Outer covering (" epithelium ectodermale " or " chrotale ") . The outer cover-
ing or chrotal epithelium (which may also be termed " ectoderma " in the more restricted
sense) in all Medusas covers the entire outer upper surface of the umbrella as a connected
dermal covering, and only passes into the endodermal epithelium at the umbrella margin
(in some Medusas at the excretory papillae of the umbrella). Corresponding to the form
of the concave-convex umbrella, we distinguish two different principal parts of its

REPORT ON THE DEEP-SEA MEDUSAE. xxix

ectodermal epithelium, which pass into one another at the umbrella margin, the dorsal and
the ventral chrotal epithelium, or the " exumbrella " and " subumbrella." Both are distin-
guished in the same way as the dorsal and ventral endoderm. The exumbrella or the
dorsal ectoderm (also termed " chrotal epithelium of the notumbrella," or shortly " upper
ectoderm ") covers the whole convex surface of the gelatinous umbrella in the form of a
delicate, flat epithelium of very uniform and indifferent character. The cells of this
epithelium are usually extremely thin, but very extensible polygonal plates, which lie
immediately on the gelatinous body, and are covered on their upper surface by a very
delicate cuticule ; this often looks dotted or granulated, as at definite distances on it
there are thickenings in the form of nodules or small papillae. Vibrating flagellate cells are
wanting for the most part on the dorsal exumbrella, but they are often found on definite
limited spaces, especially on and near the umbrella margin ; the flagella are then usually
very delicate and fine. Pigment cells and urticating cells are more frequently found in
the exumbrella, especially in the vicinity of the umbrella margin and in the projecting
radial ribs, ridges, nodes, and papillse with which the convex outer surface of the umbrella
is covered in many Medusae. The subumbrella or ventral ectoderm (also termed the
" chrotal epithelium of the coelumbrella," or shortly the " lower ectoderm ") covers the
entire concave surface of the gelatinous umbrella from the oral margin to the umbrella
margin (in the Craspedotae to the free margin of the velum, in the Acraspedae to the free
margin of the lobe-corona or of the velarium). It shows a much more complicated and
varied nature than the dorsal ectoderm. Its cells are usually higher, more cubical, partly
covered by a cuticle, partly not covered. Part of the cells bear vibrating flagella at
definite points, and feeling bristles at others. In the same way, urticating cells,
glandular cells, and also often pigment cells are richly developed in many regions. The
differentiation of the ectodermal epithelium is most varied and important at the actual
margin of the umbrella, and its appendages, such as the tentacles, marginal lobes, and
organs which are developed there. Thus we can often separate a special " subepithelial
layer of cells " from the true epithelium (which only covers the free surface). As numbers
of thread-cells are often developed, a special " urticating tissue " (" tela urticaria ") often
arises, which, especially in the Trachomedusa? and Narcomedusae, forms a thick " urticating
ring " and the " umbrella-clasps " (" peronia ") running out from it. The firm and thickly
accumulated nematocysts there lose their original function of protective weapons, and
attain the significance of a supporting dermal skeleton (urticating skeleton, § 71).
Another portion of the ventral ectoderm, and, in fact, that portion which covers the
subumbral wall of the gastrovascular system in the Craspedotae, furnishes a subepithelial
layer from which the reproductive cells, both male and female, originate in this section of
the Medusae.

§ 47. The inner covering (" epithelium, endodermale," or " gastrale "). The inner
covering or gastral epithelium (also termed "endoderma" in the more limited sense), in all

x.\ : x THE VOYAGE OF H.M.S. CHALLENGER.

Medusae, lines the entire extent of the hollow space of the gastrovascular system, consists
everywhere of a simple layer of flagellate cells, and passes into the ectodermal epithelium
only at the oral margin. Corresponding to the shape of the umbrella enclosing the gastral
space, we can also distinguish in this endoderm two different principal parts contiguous
at the umbrella margin, the dorsal epithelium, and the ventral epithelium of the gastro-
vascular system. Both show striking and constant differences. The dorsal endoderm (or
the " gastral epithelium of the notumbrella," often also simply termed the " umbral or
upper endoderm ") lines the concave inner surface of the thick dorsal umbrella, and covers
its gelatinous body in the form of a thin, uniform, flat epithelium of very indifferent
character (PL IV. figs. 6-8, du; PL IX. figs. 5-7, da; PL XXV. figs. 8-10, du). The
ventral endoderm (or the " gastral epithelium of the coelumbrella," often also simply
termed the " subumbral or lower endoderm ") covers the convex inner surface of the thin
ventral umbrella, and is stretched across its subumbral supporting plate in the form of a
high differentiated cylindrical epithelium (PL V. figs. 6-8, div ; PL IX. figs. 5-7, dw ;
PL XXV. figs. 8-10, dw). Its cells are much larger than those of the dorsal endoderm,
are often extremely high, and enclose plasma products of various kinds, fat, granules of
pigment, crystals, amyloid granules, and other products of a vital change of tissue, but
also numerous vacuoles which not unfrequently coalesce. In many places, that is at
the oesophagus, one part of these ventral endoderm cells is transformed into glandular
cells, and another into urticating cells ; epithelial muscular cells and even perhaps sense
cells appear to originate from it here in some places. Finally, it is also these ventral
endodermal cells which form the reproductive cells in all Acraspedse ; both ova cells and
sperm cells proceed from a subepithelial layer of the ventral endoderm. This ventral
gastral epithelium is plainly of the highest significance for the aggregate changes of tissue
of the Medusas, whilst the opposite indifferent dorsal endoderm is only of slight importance ;
these cells are, moreover, " flagellate cells," as in both cases they invariably bear a vibrat-
ing flagellum. This flagellum is only missing on the cathammata, those important points
at which the dorsal and the ventral endoderm are fused together. Whilst the whole
gastral cavity originally shows in the polyps a perfectly simple cup-shaped cavity with-
out radial sections, in the Medusas it is divided in the course of development into peri-
pheric radial sections, by the fusion of the two walls of the gastral cavity (the dorsal outer
wall and ventral inner wall) in definite radia. In this way there originate the important
fused plates or cathammata which represent the septa of the radial chambers. Each
" cathamma " or " septum," therefore, actually consists of two layers of the gastral
endodermal epithelium which have been laid firmly one upon the other, and fused
together at those points. These two closely connected layers can sometimes be
plainly distinguished (as in many Acraspedse, PL XXV. figs. 8-10), and are sometimes
fused into a single simple layer (as in most Craspedotas). In both cases we designate
this simple or double layer of cells as the fused plate, or cathamma plate (" lamina

liEPOItT ON THE DEEP-SEA MEDUSAE. xxxi

cathamrnalis," otherwise also " endodermal lamella," and " vascular plate"; comp. under
" Cathamma," §§ 100, 101).

§ 48. Connective tissue ("tela connectiva"). The connective tissue (padding tissue or
supporting tissue), whose various modifications are included in the idea of the " con-
nectivum"), appears among the Medusae in two different principal forms, as supporting
plates without cells ("fulcra"), and as padding tissue containing cells (" maltha"). The
two forms correspond to different phylogenetic stages of development, as the cell-less
supporting plate or fulcral plate only represents a simple secretion of the epithelium,
which has no independent value as mesoderm ; we can only consider as mesoderm the
cellular filling tissue or malthar plate, by which cells are produced from the endoderm
and are divided by an intersubstance (" secreted tissue "). Both forms are produced
in general from the endoderm ; though both forms are also produced in a few isolated
places from the ectoderm.

§ 49. Supporting tissue or cell-less connective tissue (" fulcrum," " tela fulcralis,"
"lamina fulcralis"). Under this name we include all forms of the connective tissue which
do not contain cells, and are therefore merely structureless or fibrous secretions of the
epithelia. They appear in two principal forms, which, however, appear inseparably
connected by transitions, as thin elastic membranes and as thick gelatinous masses. The
thin elastic supporting membranes are found everywhere in the bodies of the Medusae as
the foundation of the epithelium, and especially of the endoderm, although strongly
developed supporting plates often appear also among the ectoderm at definite spots, e.g.,
in the velum of the Craspedotas (PI. VI. figs. 13, 14, zv; PI. IX. fig. 7, zr) ; and in the
tentacles of many Acraspedae. The structureless fulcral lamellae are generally very thin but
very firm ; in transverse section, under strong magnifying power, they appear sometimes
simple, sometimes doubly contoured, usually strongly refractive. On account of then-
great elasticity, they are often of physiological importance as antagonists of the muscles
(e.g., as extensors of the tentacles and oral styles). This is also the case with the thick
cell-less gelatinous tissue, which is only distinguished from the thin elastic supporting
membranes by its more extensive, often very apparent volume of development. This
forms the principal mass of the gelatinous umbrella (and consequently of the whole
body) in the majority of the Craspedotae, as in this section the collosoma is usually
without cells, and appears as a structureless secretion of the endoderm ; it also forms
the principal body mass in part of the Acraspedae (e.g., Cubomedusse, Pelagidte,
Cyaneidas). The cell-less " fulcral gelatinous tissue," is, moreover, usually (or always?)
traversed by numerous elastic fibres, in the same way as the cellular "malthar gela-
tinous tissue."

§ 50. Padding tissue or cellular connective tissue (" maltha," "rete malthare," "lamina
maltharis "). Under this name (for want of a better) we include all the different forms of
cellular connective tissue, in contradistinction to the cell-less supporting plate, the fulcral

xxxii THE VOYAGE OF H.M.S. CHALLENGER.

tissue. All the different forms of the " cellular connective tissue " in the higher animals
(bones, cartilage, vascular tissue, mucous tissue, &c), belong to this filling tissue or
malthar tissue ; in the Medusae it actually appears only in two essentially different forms,
as gelatinous tissue and as chorda! tissue. The cellular gelatinous tissue ("telagela-
tinosa ") is the more important as to extent and distribution. It forms the principal mass
of the gelatinous umbrella (and therefore of the whole body) in the majority of the Acras-
pedss (namely, in most of the larger forms), whilst it is replaced by the cell-less "fulcral
gelatinous tissue " in the majority of the Craspedotae. The cells of the " malthar gela-
tinous tissue " are usually scattered sparsely at great distances in the structureless inter-
cellular substance, but sometimes also in greater numbers (namely, near the cathamma)
(PI. XXV. fig. 10). They usually proceed from the endoderm, from whose epithelial
layer they have passed into the underlying fulcral layer (" endodermal secreted layer,"
principally in the umbrella and subumbrella). Similar " ectodermal secreted tissue,
" whose cells proceed from the ectoderm (as in the velum of the Pectyllidae, PI. V. fig.
7, x ; PI. VI. fig. 13, as), are more rarely found. The consistency of the gelatinous tissue
varies greatly, as on the one hand it may become extremely soft mucous tissue (e.g., the
umbrella of the Aurelia), and on the other, a very firm, hard fibrous cartilage (e.g., the
cathamma of the Peromedusae (PI. XXV. figs. 8, 10). Near these firm fused ridges, the
cathammal plates in particular, the gelatinous tissue of many Acraspedae acquires a nature
which so resembles the true " fibrous cartilage " of the vertebrates both in histological
structure and physical quality as to be easily confounded with it. In this case the extra-*
ordinary firmness of the cellular tissue is chiefly formed by thickening and by the fibrous
differentiation of the intercellular substance, whilst the softer or firmer nature of the gela-
tinous tissue seems usually dependent upon the qualitative and quantitative development
of the elastic fibres in it. The latter comport themselves in the cellular gelatinous tissue
in the same way as in the cell-less tissue and usually pass from the ectodermal on to the
endodermal surface of the gelatinous umbrella (PL IX. figs. 5-7, itf). They are either
simple or branched, usually cylindrical, more rarely flattened like a ribbon (PI. VI. fig.
19). They are sometimes combined into an elastic network or grouped in branches (PI. V.
fig. 8, us). The second principal form of the padding tissue is the characteristic chorda!
tissue (" tela chordalis "), which greatly resembles the tissue of the " chorda dorsalis " of
vertebrates. It is found everywhere in the solid tentacles of the Medusae, and forms
their characteristic firm axis. This is usually cylindrical, and consists of a single row of
large, discoid, flat, circular endodermal cells, lying one above the other like the coins in
a rouleau of sovereigns (PI. I. fig. 57 ; PL VI. fig. 17; PL XII. fig. 11 ; PL XIII. figs.
5,6). Each cell is surrounded by a very thick, firm, elastic membrane, and encloses con-
tents as clear as water. The protoplasma of the cell is usually limited to a thin wall-
layer, lining the inside of the capsule-shaped membrane, and to a central axial cord, which
connects the middle of the proximal and distal wall-layer; the two are sometimes connected

REPORT ON THE DEEP-SEA MEDUSAE. xxxiii

by a network of fine threads of protoplasm, traversing the cavity of the cell (PI. I. fig. 7).
The nucleus sometimes lies in the middle of the axial cord, sometimes at the one end.
These chordal cells belong to the largest cells of the body of the Medusa ; they are
often visible to the naked eye, being about 1 mm. broad (p. 38). They sometimes also
form a special chordal ring at the umbrella margin (PI. VIII. fig. 8, y). In the short
thick tentacles, the chordal cells of the tentacle axis seem sometimes disposed in layers
(PI. IV. figs. 5-8, yt ; PI. VI. figs. 12-15, clt). The base of the axis is usually still continu-
ously connected with the endoderm of the coronal canal ; more rarely it becomes com-
pletely separated from it, and, therefore, mesodermal (Pis. IV., VI., XII., &c). In the
gelatinous tissue the small cells retreat entirely against the powerful intercellular sub-
stance ; the reverse is the case in the chordal tissue.

§ 51. Muscular tissue (" tela muscularis "). The muscles of all Medusas consist of fine
muscular fibiillas laid parallel, which are connected somewhere with a small lump of proto-
plasm containing nuclei, and must, therefore, be regarded as filamental processes of mus-
cular cells. The fibrillas are usually very long and thin, sometimes cylindrical, sometimes
flattened like a ribbon ; in most longitudinal or radial muscles the fibrillas are smooth, not
striated, but they are more or less plainly striated in the transversal or circular muscles.
Both the smooth and the striated muscular cells originate for the most part from the
ectoderm. In isolated spots, however, both kinds are also formed from the endoderm (as
for example at the oesophagus and the oral arms). With regard to the relation of the
muscular cells to their original place of formation, the epithelium, we distinguish two
principal forms of the muscular tissue ; epithelial muscular cells and mesodermal muscular
cells ; the latter still lie in the true epithelial layer of the upper surface or immediately
below it, whdst the latter have become completely separated from it and form an inde-
pendent, though thin, mesodermal layer.

§ 52. Epithelial muscular cells ("myoblasti epitheliales," "tela muscularis epitheli-
alis "). The majority of the Medusas muscles, that is in the section of the Craspedotas,
are composed of smooth or striated fibrillar, whose muscular cells do not form an inde-
pendent mesodermal layer, but either belong to the endodermal epithelium itself or to a
subepithelial layer lying immediately below it. The fibrillas of these " epithelial muscular
cells " or " neuro-muscular cells " therefore lie immediately under the epithelium from
which they proceed, and on the supporting plates on which they are borne. They are
usually placed in a parallel layer beside each other, or arranged in several layers one above
the other in such a way that they form flat leaves or lamellas. By further develop-
ment of the muscles, these " muscular leaves " become arranged in folds, whilst the sup-
porting laruellas bearing them forms corresponding composite folds by local thickening,
as, e.g., in the larger hollow tentacles of the Geryonidas, and of the Cyaneidas, &c. ; in
the velum of some Craspedotse (PI. VI. figs. 13, 14). In the smaller and lower
Medusas, the broad coronal muscle forms a simple, smooth, band-shaped plate on

(ZOOL. CHALL. EXP. — PART XII. — 1881.) M e

xxxiv THE VOYAGE OF H.M.S. CHALLENGER.

the umbrella surface, whilst in many larger and higher Medusae the supporting
plate of the subumbrella is raised in concentric circular folds, which are covered by
corresponding folds of the muscular plate (e.g., Lucernaria, PL XVII. fig. 20 ; Peri-
phylla, PI. XXII. fig. 22).

§ 53. Mesodermal muscular cells (" myoblasti mesodermales " " tela muscularis meso-
dermalis "). When the folded epithelial muscular leaves increase in extent and become
further developed, their growth is not limited to the formation of folds, but the epithelial
or subepithelial muscular cells emerge completely from their point of origin, the epithe-
lium, and form independent " mesodermal muscular cells." As they separate in great
quantities from the epithelium and become united to special plates or bundles, they pass
inwards in the connective tissue and form perfectly independent mesodermal muscles.
Such mesodermal muscles are more commonly found in the system of the longitudinal
muscles than of the circular muscles, more rarely and chiefly in the larger species among
the Craspedotse, but more frequently among the Acraspedas. Thus, for example, in the
large Peromedusas, the powerful deltoid muscles of the subumbrella, the longitudinal
muscles, and root muscles of the tentacles, &c. (Pis. XX.-XXIV. md, mh, &c), belong
to this category. Here the muscles are frequently detached so completely from the
epithelium that later on they are separated from it by a special supporting lamella or
even a thick gelatinous plate. The external coronal muscles of Atolla, which increases to
4 mm. broad by 2 mm. thick, and are composed of many layers of coronal muscular fibres,
lying the one above the other, form one of the strongest mesodermal masses of flesh (PI.
XXIX. figs. 4, 7, 8, one"). Among the Craspedotaa, Pedis furnishes an example of
strong mesodermal muscles in the velum and subumbrella (PI. V. fig. 7 ; PI. VI. figs.
12-14, on).

§54. Nervous tissue (" tela nervea.") The two essential component parts of nervous
tissue, which are distinguished as nerve cells and nerve fibrillar in the higher animals,
are also already differentiated in the Medusae ; the two compose the central as well as
the peripheric part of the nervous system, even though their situation and connection
in the central nerve ring of the umbrella margin is different from that in the nervous
plexus of the subumbrella, and so forth. Both the nerve cells and the nerve fibrillse,
which proceed from and connect them, are, for the most part, products of the
ectoderm and have hitherto been considered to be exclusively such. Nerve cells and
nerve fibres are found in isolated places in some (and perhaps all) Medusae, which
originate from the endoderm, as on the very movable and sensitive oral parts (oral
filaments, oral lobes, oral arms, oral pouches). It is probable that on the gastral inner
surface of these oral organs the endoderm forms both muscular cells and nerve cells,
among the latter we may perhaps look for specific sense cells (gustatory cells ? olfactory
cells ?). These difficult conditions require much closer investigation. In any case,
both this localised and slightly extended ectodermal nervous tissue, and the more

REPORT ON THE DEEP-SEA MEDUSAE. XXXV

important and widely diffused ectodermal nervous tissue are most closely connected with
the epithelium from which they are originated. The whole muscular system in the
Medusse preserves more or less its original epithelial character. We can therefore
distinguish here, as in the muscular tissue, purely epithelial and subepithelial cells ; the
former may usually be termed sense cells, the latter ganglion cells. Both are connected
by very fine fibrillse, thread-shaped processes, which are repeatedly branched and com-
bined into network and nerve plexus.

§ 55. Sense cells or epithelial nerve cells ("cellulse sensillares," "sensoblasti"). Under
this title we include all these nerve cells (in the widest sense) which lie in the
epithelium itself, and which have fine thread-shaped processes or fibrillar at their bases,
by which they are connected with other cells of the nervous system. These sense cells are
sometimes scattered singly in the epithelium between its indifferent covering cells {e.g.,
on many places in the tentacles and oesophagus), sometimes, as a connected covering,
they form a true sensitive epithelium {e.g., on the dorsal nerve ring of the Craspedotae,
and on the margin of the velarium of the Acraspedae). We can distinguish two
principal forms of sense cells, indifferent or neutral, and differentiated or specific sense
cells. We call those epithelial sense cells " indifferent," to which we cannot assign a
specific function of sense, and which therefore represent the oldest and simplest form of
the nervous elements. Provisionally we may consider as such in the Medusas, all those
flagellate cells and bristle cells of the ectoderm which are connected at their base,
directly or indirectly, by processes and nerve fibrillar with other nerve cells, and in which
we recognise no specific sensitive function (PI. XIV. fig. 9). All these neutral sense cells
have a fine hair-shaped process on their free upper surface, which is movable in the
sensitive flagellate cells {e.g., on the sense epithelium of the nerve ring), but stiff in the
bristle cells. We include in the latter the true tactile cells (without nematocysts, with a
tactile bristle, " palpocilium ") and also the thread cells (with nematocysts and with an
urticating bristle, " cnidocilium "). How far the flagellate cells and thread cells of the
ectoderm belong to the category of indifferent sense cells depends upon their passing at
their base into communicating processes or nerve fibrilke (comp. below § 79, organs of
touch). We may consider as differentiated or specific sense cells such epithelial nerve
cells to which from their situation, structure, or connection, we can assign some specific
sensitive function. To this category belong — (l) the olfactory cells (or gustatory cells ?)
on the clavelhe of the Craspedotas (PI. II. fig. 8, q) and on the protective scales of the
rhopalia in the Acraspedse ; (2) the vision cells of the eyes, which are sometimes
differentiated into pigment cells and rod cells ; (3) the auditory cells of the auditory
organs (PL VI. fig. 16). The last bear a free, usually long and thin auditory hair, and
therefore do not differ in form from the ordinary tactile cells (with a tactile bristle) from
which they are also derived phylogenetically ; they become " auditory cells" as the)" are
in functional connection with " otolite cells " which contain an otolite, and belong to the

xxxvi THE VOYAGE OF H.M.S. CHALLENGER.

ectoderm in one order only (the Leptomedusaj) and in all other orders to the endoderm.
In some cases {e.g., in most Leptomedusse and in the Geryonidse) the auditory cells
become completely separated from the free epithelium to which they originally belonged ;
they are then transformed into mesodermal interior epithelium, as the open olfactory
depressions become detached from the dermal covering and from closed auditory
vesicles.

§ 56. Ganglion cells or mesodermal nerve cells (" celluke gangliosse," " neuroblast! ").
The ganglion cells bear the same relation to the sense cells as the mesodermal muscular
cells do to the epithelial. The ganglion cells are, in fact, subepithelial nerve cells
secreted from the epithelium, from which they have originated both ontogenetically and
phylogenetically ; they are still connected directly or indirectly with this their point
of origin by thread-shaped processes, the nerve-fibrillae (PI. XIV. fig. 10). All gang
lion cells of the Medusae appear to have two or more processes, and are, therefore,
either bipolar (fusiform) or multipolar (stellate) cells. Both forms appear both in the
central and in the peripheric nervous system ; the bipolar cells, however, preponderating
in the central nerve ring, the multipolar ganglion cells in the peripheric nervous
plexus ; the former therefore lie principally in the umbrella margin, the latter in the
subumbrella. The central ganglion cells, moreover, both in the nerve ring and in the
organs of sense, show definite conditions of relation and position to the neighbour-
ing organs, especially to the sense cells of the epithelium. In the Craspedotse, the dorsal
(or exumbral) nerve ring covered by the sense epithelium is formed for the most part
of parallel lying, circular fibrillae, and is much poorer in ganglion cells than the ventral
(or subumbral) nerve ring which has no sense epithelium and is more motor. In the
Acraspedae the ganglion cells seem rather to be accumulated at the bases of the sense
clubs, and to form four or eight ganglia which are sometimes connected by a centralised
ring of bundles of fibrillse (Cubomedusae), sometimes by a more decentralised plexus
of fibrillae. The peripheric ganglion cells are scattered, sometimes sparsely, sometimes
pretty numerously in the nervous plexus, which extends chiefly in the subumbrella in
the form of delicate,reticulate plexus of fibrillae ; this subumbral plexus lies between
the muscular plate of the subumbrella and the endodermal epithelium, from which the
latter has arisen. Both this peripheric nerve plexus and the central nerve ring may
already be regarded as mesodermal nerves, as they possess independent cells, secreted
from the epithelium.

REPORT ON THE DEEP-SEA MEDUSAE. XXXTli

III NEUEODERMAL SYSTEM OF THE MEDUSAE.

§ 57. Composition of the neurodermal system. Of the large organic systems compos-
ing the body of the Medusae, the neurodermal system includes the aggregrate animal
organs, the apparatus of sensation and motion. This is therefore opposed physiologically
to the gastrovascular system, which forms the complex of the vegetative organs. This
antithesis is also shown histologically in reference to the two primary germinal layers, as
the greater and most important parts of the neurodermal system originate from the ecto-
derm (or "animal germinal layer"), whilst those of the gastrovascular system originate
chiefly from the endoderm (or "vegetative germinal layer"). The apparatus of motion,
formed by the umbrella and the wide-spread muscular plates, situated on the concave
surface of the umbrella cavity, is by far the more considerable although the less differen-
tiated of the two apparatuses which compose the neurodermal system. The apparatus of
sensation is less extensive, but more strongly differentiated ; it is situated chiefly on the
umbrella margin, and includes the nervous system along with the tentacles and differenti-
ated organs of sense.

§ 58. Umbrella (u). The typical and most characteristic principal organ of the Medusae,
which distinguishes them from the ancestrally-allied polyps, is their peculiar swimming
organ, the umbrella. From its volume and weight this always forms the principal mass
of the body, and consists of a voluminous gelatinous body (" collosoma ") which contains a
large amount of water, and is sometimes almost as soft as mucus, sometimes almost as
hard as cartilage. It is more or less " umbrella-shaped," convex above, arched concavely
below. The general form of the umbrella, however, varies greatly. Sometimes its vertical
diameter (or "the central principal axis") is greater than the greatest horizontal diameter
(or the transverse axis), and the umbrella is thin, conical, bell-shaped, pyramidal, or
obelisk-shaped (as in most Anthomedusse, Stauromedusae, Peromedusae, and Cubomedusae,
Pis. XV.-XXVL). Sometimes, on the contrary, the horizontal diameter is greater than the
vertical, and the umbrella, therefore, more discoid, hourglass-shaped, or semi-spheroidal
(as in most Leptomedusae, Trachomedusae, Narcomedusse and Discomedusaa (Pis. I.-XIV.,
XXVII. -XXXIL). The gelatinous body is usually thickest in the middle of the umbrella,
and decreases sometimes regularly, sometimes more suddenly towards the umbrella
margin. If we take the umbrella and the parts of the gastrovascular system enclosed in it
as a whole, we may term the outer convex surface the "outer umbrella" or "exumbrella"
(<>), and the inner concave surface the "inner umbrella" or "subumbrella" (w). More
accurately speaking, however, the umbrella consists of two distinct gelatinous plates which
may be distinguished as the dorsal umbrella ("umbrella dorsalis," or " notumbrella") and
the ventral umbrella ("umbrella ventrabs," or "coelumbrella"); the former corresponds
to the "calyx" or dorsal wall of the polyps, the latter to their " peristomium " or ventral
wall. Both walls pass immediately one into the other only at the umbrella margin, and

xxxviii THE VOYAGE OF H.M.S. CHALLENGED

arc elsewhere separated by the hollow space of the gastrovascular system, whose bounding
surfaces are only fused together at definite points. The exumbrella is the free convex
surface of the thick dorsal wall ; the subumbrella is the free concave surface of the thin
ventral wall.

§ 59. Umbrella dorsalis (" notumbrella," upper or dorsal umbrella). In the more
limited sense this part is usually simply termed the " umbrella," as it forms the principal
mass of the umbrella, and as its voluminous gelatinous disk is much thicker than the thin
gelatinous plate of the ventral umbrella. Its upper convex surface, covered with dorsal
ectoderm, is the "exumbrella" (c). Its lower concave surface forms the outer or
abaxial wall (umbral wall) of the gastrovascular system, and is covered by its flat " dorsal
endoderm." The two epithelial layers of the dorsal umbrella, the outer ectodermal layer
and the inner endodermal, are separated by the powerful mass of the gelatinous body
(" collosoma "). They never run into one another, as they pass immediately at the
umbrella margin into the two corresponding epithelial plates of the ventral umbrella.
The endodermal epithelium of the dorsal umbrella consists of flagellate cells of an
indifferent nature, whilst its ectodermal epithelium often forms thread cells, more rarely
also epithelial muscular cells (" exumbral muscles ").

§ 60. Exumbrella. The convex outer surface of the dorsal umbrella, which we call
shortly the " exumbrella " in many Medusa?, is perfectly smooth, arched equally without
any special characteristic, and covered uniformly by the simple ectodermal epithelium. In
many other Medusa?, on the contrary, it is distinguished by repeated projections in the
form of nodes, ribs, ridges, spicules, &c. These projections are often distinguished by
accumulations of thread cells, often also of pigment cells, and therefore serve as weapons of
defence of the umbrella. Projecting radial urticating ribs are found among the Craspe-
dota? that is in many Anthomedusa?, e.g., four perradial in several Codonida? and Tiarida?,
eight adradial in Ectopleura and Ctenaria (System, fcaf. vii. fig. 7), sixteen in Pec-
tyllis (Pis. III., IV.), Pectanthis (Pis. VII., VIII.), and Tesscrantha (PI. XV). In
Corynetes the whole exumbrella is overspun with a network of ridges, having projecting
urticating papilla? at their points of junction ; more commonly the urticating papillae
are scattered equally over it (e.g., Thamnostylus, PI. I.). Among the Acraspeda?,
such projecting urticating ribs are of less morphological importance than the deep furrows
of the exumbrella, by which the latter is divided into a number of gelatinous plates.
Very often there is a deep and distinct circular furrow or coronal furrow, which separates
the central umbrella disk from the peripheric umbrella corona. Whilst the former is usually
smooth, the latter, on the contrary, is often divided by radial furrows into projecting "gela-
tinous sockels or pedalia," which serve to bear the tentacles and rhopalia, as e.g., in Peri-
phylla (Pis. XVIIL, XIX.), Nav.phanta (Pis. XXVII., XXVIIL), and Atolla (PI. XXIX.).
The exumbrella of C&phea (System, taf. xxxii.) bears large, conical urticating papilla?.

§ Gl. Umbrella ventralis (" eoelumbrella," lower umbrella or ventral umbrella).

REPORT ON THE DEEP-SEA MEDI7SJE. xxxix

This part of the umbrella is usually simply termed umbrella iu the wider sense, though
this name really belongs only to its lower concave surface, which is covered by ectoderm.
Its upper, convex surface forms the inner or axial wall (subumbral wall) of the gastro-
vascular system, and is covered by its high " ventral endoderm." The two epithelial
layers of the ventral umbrella, the ectoderm of the concave surface (subumbrella), and the
endoderm of the convex surface, only pass into one another at the umbrella margin, and
are likewise separated by a thin but firm supporting plate (210). This fulcral lamella of
the subumbrella is equivalent to the thick gelatinous body of the dorsal umbrella,
though much thinner, and passes immediately into the umbrella margin at the edge of the
latter. The endodermal epithelium of the ventral umbrella consists of high flagellate
cells, which also often form glandular cells, whilst its ectodermal epithelium (the " sub-
umbrella " in the more restricted sense) originates the most important part of the mus-
cular system of the Medusas (" subumbral muscles").

§ 62. Subumbrella. The convex inner surface of the ventral umbrella, which we
designate shortly as the " subumbrella " (in the more restricted and special sense), is of
special importance as bearing the muscular system, which affects the swimming motions
of the Medusas. It is, moreover, distinguished by varied differentiations of the ectodermal
epithelium, lining the umbrella cavity enclosed by it. Thus, for example, glandular cells,
pigment cells, and thread cells are often found disposed in a certain order on its ectoderm,
and in all Craspedotas it also forms the point of origin of the reproductive glands.
Whilst in most of the smaller Medusas (for example, the Craspedotas) the subumbrella
appears smooth and regularly vaulted, in most of the larger Medusas (chiefly Acraspedas)
it is folded repeatedly and distinguished by special projections. Among the more im-
portant of these are the gelatinous ridges which serve for the wider superficial extension
of the subumbral muscular system. They usually run in concentric rings {e.g., on the
coronal muscle of the Peromedusas (Pis. XIX., XXII.), more rarely in radial bunches (e.g.,
Drymonema (Pis. XXX., XXXI. ). The mesenteries of many Anthomedusas, and Tracho-
medusas, Stauromedusas, and Cubomedusas may be mentioned as special processes of the
subumbrella, which project into the umbrella cavity in the form of vertical radial septa ;
we shall speak of them further on in the "umbrella cavity" along with the various
secondary cavities and niches, which penetrate from the umbrella cavity into the sub-
umbrella (§§91, 94).

§ 63. Central and peripheric umbrella (" discus centralis " and " corona peripherica ").
In all Medusas a certain morphological and physiological contrast can be recognised more
or less distinctly between the central and the peripheric part of the umbrella ; the most
important part of the vegetative gastrovascular system lies in the former, the most
important part of the animal neurodermal system in the latter. We term the central
principal section of the umbrella, enclosing the stomach and mouth along with the oral
organs, the umbrella disk (" discus umbralis "), the peripheric principal section containing

xl THE VOYAGE OF H.M.S. CHALLENGER.

the umbrella margin along with the most important part of the muscular and nervous
systems (sense organs and tentacles) the umbrella corona (" corona umbralis "). These
two principal sections of the umbrella correspond at the same time to the two principal
sections of the gastrovascular system, as the central principal intestine is situated in
the umbrella disk, but the peripheric coronal intestine in the umbrella corona. The
boundary between the disk and the corona is often sharply defined externally, as an
exumbral coronal furrow (" fossa coronaris ") is inserted more or less deeply between the
two, as in many Narcomedusae (Pis. IX.-XIL), Peromedusas (PI. XVIIL), Cuboinedusse (PI.
XXVI.), and in a few Discornedusse, very distinctly in many Cannostomaa (Pis. XXVIL-
XXIX.). The central umbrella disk is more discoid or lens shaped (" umbrella lens,"
" lens umbralis") in the depressed Medusa}, but more conical or bell shaped (" umbrella
cone " " conus umbralis ") in the higher vaulted Medusae. In the Craspedotas the
peripheric umbrella corona ends in the typical velum of this section, but in the
Acraspecla3 in the characteristic lobe corona or velarium.

§64. Umbrella peduncle and umbrella cupola ("pedunculus umbralis" and " cupola
umbrellse "). In many Medusa?, though only in the minority, the " apex of the umbrella "
(the uppermost, aboral and proximal part of the " notumbrella ") is not arched and
rounded as usual, but prolonged into a projecting apical process or a conical, peduncle-
like process. In one order only, the Stauromedusse, it is developed into a true umbrella
peduncle (" pedunculus umbrellse "), whose aboral end, the " foot plate," serves for adhesion
to the bottom of the sea or to foreign bodies (Pis. XVI., XVII. ; System, taf. xxi., xxii.).
But in many other Medusae of different orders (namely, of Anthomedusae and Peromedusse)
in place of an apical peduncle we find a peculiar umbrella cupola or conical apical
process (" cupola umbrellas ") at the top of the umbrella. This is the ecpiivalent of the
adhering peduncle, and contains, like it, a caecal axial "apical canal" or "peduncle
canal" (Tesserantha, PI. XV. figs. 1-3, p). A special morphological interest attaches
itself to these parts in that they are heirlooms from the polyp ancestors of the Medusse,
and are homologous with the peduncle and peduncle canal of the polyps, by means of
which the latter are fastened to the bottom of the sea.

§ 65. Gastral peduncle and gastral cone (" pedunculus gastralis " and " conus gastralis ").
An oral process is often found, though only in the Craspedotae, on the concave inner
surface of the central umbrella disk, in the same way as the umbrella peduncle or the
umbrella cupola is developed from it as an aboral process. This oral process first appears
as a flat, insignificant, conical elevation in the centre of the endodermal hollow surface
of the " notumbrella," and projects, more or less, into the central gastral cavity (System,
taf. xi. xiv. xv.). By further growth from this gastral cone, a long cylindrical
gelatinous peduncle is developed, which projects far into the umbrella cavity or even
beyond the umbrella opening, and which takes the surrounding parts of the " coel-
umbrella" along with it. The gastral sac lies no longer, as usual, in the bottom of the

REPORT ON THE DEEP-SEA MEDUSAE. xli

umbrella cavity, but at the distal end of a free solid gastral peduncle. The radial canals
originating; in the bottom of the stomach run in the ectodermal outer surface of the
cylindrical gastral peduncle (which is often also quadrangularly prismatic, pyramidal, or
conical) to the bottom of the umbrella cavity, turn over on to the subumbrella, and run
in it to the umbrella margin. The longitudinal muscles which move the gastral peduncle,
alternate with its ascending radial canals. The solid gastral peduncle frequently
resembles the hollow, and likewise proboscis-shaped oesophagus of the Craspedotse and
has often been confounded with it. The gastral peduncle is never found among the
Acraspedse, but is very frequent in all four orders of the Craspedotae. It is most
strongly developed in a part of the LeptomedusEe (Saphenidse) and Trachomedusse

Fig. C. Odorchis germanica (Leptomedusa;, Eucopidai).

Profile view, (ug) Gelatinous umbrella, (us) Solid gelatinous peduncle of the stomach, (v) Velum.
(cm) Velar marginal vesicles, (tp) Perradial tentacles, (ti) Interradial tentacles, (s1) Distal testis
(on the subumbrella). (s-) Proximal testis (on the oesophagus). (cj>) Perradial canals, (g) Stomach.
(al) Oral lobes.

(Geryonicke). (Comp. System, taf. iv. xh. xiii. xviii. xx.). In a portion of the
Geryonidse it runs out still further below, past the central part of the umbrella peduncle
and forms a pointed cone, projecting freely like a tongue into the hollow space of
the stomach, which is fastened below to the gastral peduncle (System, taf. xviii. fig. 5).
This tongue-like cone (" conus lingualis," " glossoconus ") is perhaps an organ of taste.

§66. Umbrella margin ("margo umbrellae," um). The umbrella margin forms the
lower or distal boundary line of the umbrella, at which its two walls, the dorsal and the
ventral wall, pass into one another ; at the same time the exumbral epithelium of the
convex dorsal umbrella (qe) proceeds directly at this boundary line into the subumbral

(ZOOL. CHALL. EXP. PART XII. — 1881.) M /

xlii THE VOYAGE OF H.M.S. CHALLENGER.

epithelium of the concave ventral umbrella (qw). The umbrella margin is the most
important part of the neurodermal system in all Medusae, both morphologically and
physiologically, as in it the most important animal organs — organs of sense, nerves and
muscles — attain their highest development. The central part of the nervous system and
the tentacles especially are always originally situated in the umbrella margin. The
umbrella margin is also of great importance for classification, as it is chiefly on it that
the variations of formation appear which lead to the distinction of genera and species.
In fact the distinction and nomenclature of the two principal divisions of the Class
Medusae, of the two sections Craspedotae and Acraspedas, are taken from the umbrella
margin, which presents important and striking diversities in the two sections. The
" velum " is characteristic of the former, the " lobe corona" of the latter.

§ 67. Umbrella margin of the Craspedotse: velum ("diaphragma"). In all Craspedotae
or Hydromedusae a direct process of the free umbrella margin projects inwards from it ;
the marginal veil or "velum" (also termed "swimming membrane" or "diaphragma")
is wanting in all Acraspedae or Scyphomedusae. The velum forms a thin, membranous,
broader or narrower ring, which in a state of rest sometimes hangs loose vertically from
the umbrella margin, and is sometimes stiffly stretched horizontally and projecting inwards,
narrowing the entrance of the umbrella cavity more or less. In the Pectyllidas (Pis. III.—
VIII.) the velum is so broad that it Can probably close the entrance into the umbrella
cavity when fully extended. In most Narcomedusae it is very broad, whilst it is very narrow
in many Leptomedusae ; in Obelia it is rudimentary. We can always distinguish in the
velum a free distal margin and a basal proximal margin, inserted at the umbrella margin ;
likewise a ventral inner surface and a dorsal outer surface. The ventral or suburnbral
surface of the velum is covered with the ectoderm of the " suburnbrella," the dorsal or
exumbral surface with the ectoderm of the " exumbrella " ; below the latter there lies a
thin supporting plate, below the former a muscular plate, composed of circular fibres,
which is a direct process of the coronal muscular layer of the suburnbrella (comp. Pis.
IV. -VI., IX.-XIV.).

§ 68. Collar lobes of the Narcomedusaa. Whilst in most Craspedotaa the velum is
stretched at equal breadth all round the umbrella margin, in the order of the Nar-
comedusae it undergoes peculiar transformations as the umbrella corona (or " collar "), and
is separated by deeper or shallower incisions of the margin into a number of separate
" collar lobes " (at least four, Cmiarcha, PL IX. ; usually eight or more, Pis. X.-XIV.).
These often closely resemble the true marginal lobes of the Acraspedae, and are usually con-
founded with them ; they differ entirely, however, in origin, structure, and signification.
The collar lobes of the Narcomedusaa originate in the tentacles, abandoning their original
insertion on the umbrella margin and migrating more or less upwards into the umbrella.
They then take with them a process from the urticating ring of the umbrella margin, in
the form of a radial (centripetal) urticating streak, which as the "umbrella clasp"

REPORT ON THE DEEP-SEA MEDUSAE.

xliii

(" peronium ") keeps up the connection between the ectodermal epithelium of the
tentacle and the umbrella margin. The peronia transect the gelatinous body as far as the
subumbrella, and at the same time form deeper or shallower indentations at the dorsal
bases, which in the Peganthida3 (Pis. X.-XII.) become deep incisions of the umbrella
margin. The velum which fills them naturally appears much broader here than at the
margin of the underlying collar lobes ; it connects the latter in the same way as the
velarium of the Cubomedusae does the true marginal lobes of this order (§ 70).

§ 69. Umbrella margin of the Acraspedse: lobe corona (" corona lobaris"). Whilst the
velum appears in all Craspedotas or Hydromedusse as a characteristic process of the umbrella
margin, it is generally wanting in all Acraspedse or Scyphomedusae. A velum-like mem-

fig. D. Pcrkolpa quadrigata (Peromedusre, Pericolpidae).

Subumbral aspect, (oi) Sense clubs (interradial). (oa) Ampulla at their bases, (t) Tentacles (perradial).
Ifil) Marginal lobes (adradial). (bu) Horseshoe-shaped canals of the lobes. (Jet) Peronia between
the two limbs of the canals, (s) Genitalia, (ml:) Root muscles of the tentacles. (mdl) Perradial
deltoid muscles, (md-) Interradial deltoid muscles, (ak) Oral ribs, (ar) Oral grooves on their
inner surface, (ai) Oral tseniola. (ti) Tentacle roots, (be) Coronal pouches, (mc) Coronal muscles.

brane (" velarium"), which in some families of the latter (Charybdeidae, Aurelidse) forms
a narrower or broader membranous selvage at the umbrella margin (PI. XXVI. va), is
perfectly different from the true velum both as to origin and structure. On the other
hand, all Acraspecke have a lobe corona (" corona lobarum "). This is usually wanting in
the Craspedotse, as the gelatinoiis " collar lobes " which are developed in some groups of
the latter (Narcomedusge, Pis. IX.-XIV.), but are connected by the velum, cannot
be compared to the true marginal lobes of the Acraspedse. The latter may rather be
essentially considered as shallow, leaf-shaped " steering tentacles." They have one or
two longitudinal muscles on their concave subumbral surface, by whose contraction they

xliv THE VOYAGE OF H.M.S. CHALLENGER.

can operate like a helm in the motion of swimming. From the ontogeny of the
Acraspedae (Amelia) it is probable that their marginal lobes have really originated
phylogenetically from tentacles. A three-cleft Scyphostoma tentacle can have given rise
in the Ephyra formation to a rhopabum and the two " ocular lobes" enclosing it. The
number of the lobes varies greatly. In the Acraspedae there are at least eight adradial
marginal lobes (fig. D, Pis. XVL-XVIL). In place of these, however, we usually find
sixteen subradial (Pis. XVIII.-XXVIIL), and their secondary number is often con-
siderably increased (Pis. XXX.-XXXIL).

§ 70. Velarium of the Cubomedusae. Whilst in most Acraspedae the marginal lobes
project freely at the umbrella margin, alternating with the tentacles and rhopalia, the
Craspedotae are distinguished by the marginal lobes being fused together or connected by a
thin intermembrane, like a swimming membrane (PI. XXXVI. ; System, taf. xxv.-xxvi.).
In this way a muscular, broad, thin marginal membrane is formed, which strongly
resembles the velum of the Craspedotae, and has hitherto been generally considered
homologous with it : it differs completely from the latter, however, both in its origin and
its finer construction, and is therefore more appropriately termed velarium. The true
velum of the Craspedotae, and the velarium confounded with it of the Cubomedusae have
arisen quite independently of one another and in a different manner ; the two bear a
completely different morphological relation to the umbrella margin and to its nerve ring.
The velarium of the Cubomedusae is usually traversed by canals (distal processes of the
coronal pouches), (PL XXVI. fig. 8), whilst this is never the case in the velum of the
Craspedotae. Moreover, the velarium in most Cubomedusae is suspended by four
perradial "frenula" (or gelatinous ridges of the subumbrella. Comp. above and
PI. XXVI. figs. 2, 8, vf). The velarium differs in the two families of the Cubomedusae,
inasmuch as it is composed of eight adradial marginal lobes in the Charybdeidae, but of
sixteen subradial marginal lobes in the Chirodropidae. The marginal lobes are fused to
a velarium in the same way, but not so apparently in many Discomedusae, such as the
Ehizostomae. It is very broad, for example, in Drymonema (Pis. XXX., XXXL). A
narrow circular border of the umbrella margin, which in some Discomedusae is developed
below the corner of tentacles (Aurebdae, System, taf. xxxiii. fig. 8, vet) differs both from
the velarium and from the true velum of the Craspedotae.

§ 71. Urticating organs (" nematillae," " nematophora," " organa urticantia," n).
In all Medusae, as in Acalephae or Cnidaria in general, special organs are formed from
the epithelium at definite parts of the body, which are essentially composed of nema-
tocysts (" cnidoblasti"), and are therefore termed urticating organs or "nematillae."
In the Medusae these are for the most part products of the ectoderm, whilst the endo-
derm only forms nematillae in a few places, as for example, on the gastral filaments and
in the oral cavity. The urticating organs serve chiefly as weapons of attack and defence
(as. for example, the tentacles), but at the same time also as firm supports of the soft body

REPORT ON THE DEEP-SEA MEDUSJB. xlv

(as, for example, at the umbrella margin). The indicating weapons (" arma urticaria ")
appear under very varied forms as roundish urticating knobs (in the whole ectodermal
surface), closed urticating rings (in the outer wall of the tentacles), narrow urticating
streaks or flat urticating pads (at the umbrella margin), conical urticating papillae (in the
exumbrella and subumbrella), composite urticating clubs and urticating batteries (at the
end of the tentacles), and so forth. All these urticating weapons consist of epithelial
accumulations of numerous urticating cells, which usually lie compacted in the upper
surface of the ectoderm, and which throw the urticating threads and fluid from
their thread cells, when their freely projecting urticating bristle (" cnidocilium ") is
touched. They are usually developed in a subepithelial layer, the " interstitial tissue."
As soon as the thread cells and their filaments are fully developed in the cnidoblast they
become erect, and pass from the subepithelial into the superficial epithelial layer.
"When this is very thin and flat {e.g. in the exumbrella) the thread cells originate in the
epithelial cells of the upper surface itself. In many places, principally on the umbrella
margin, the thread cells lose their original significance as weapons, accumulate thickly
compacted in firm masses, and so assume the function of a supporting skeleton. Such
subepithelial urticating skeletons (" sceleta urticaria ") attain a high development in the
Trachylinae (Trachomedusae and Narcomedusae). They sometimes form a firm urticating
ring on the umbrella margin (on the distal margin of the coronal canal (Pis. IX.-XIV.
nc), sometimes radial urticating streaks, which run centripetally from the urticating ring
and rise upwards in the exumbrella. These centripetal urticating streaks serve as firm,
elastic support, sometimes for the freely projecting auditory clubs (auditory clasps " oto-
porpas," Pis. IX.-XIV. oo), sometimes for the dorsally inserted tentacles, whose bases they
connect with the umbrella margin (umbrella clasps, " peronia," Pis. IX.-XIV. en; comp.
§ 68). As the cnidoblasts in these supporting shields are accumulated in a number of
layers, the one above the other, and lie deep under the epithelial upper surface, the en-
closed filaments, which are no longer able to escape, lose their function as armature,
whilst the hard nematocysts which assume the supportive function of the firm and elastic
cartilaginous tissue become proportionately more strongly developed (PL XIV. fig.
12, en).

§72. Nervous system. In all Medusas the nervous system stands at a very low
stage of development, as it retains the most immediate connection with its place of de-
velopment, the ectodermal epithelium, and as neither its central nor its peripheric parts
have become completely and independently separated.- We can usually distinguish in
all Medusae a central and a peripheric section of the nervous system. The circular
central part lies either on the umbrella margin or above it on the subumbrella ; whilst
the peripheric part extends chiefly on the subumbrella in the form of a diffuse nervous
plexus. Both in the central and peripheric part we find smaller and larger ganglion cells,
mixed with finer and coarser fibrillar (PL XIV. figs. 9, 10). These cells are most closely

xlvi THE VOYAGE OF H.M.S. CHALLENGER.

connected on the one side with the " sense epithelium " of the ectoderm lying below it
(namely, at the umbrella margin and the organs of sense), and on the other with the
underlying muscular plate (namely, at the subumbrella and the oesophagus). Inde-
pendent " ganglia " separated into units, or centralised nerve knots, and visible " nerve
fibres " consisting of bundles of separate nerve fibrdlae are only developed in a few sur-
faces {e.g., Charybdea, PL XXVI). We ought, however, to observe that the most recent
numerous and important researches on these difficult conditions are still too insufficient
to allow us to form exhaustive and certain conclusions on the subject. On the one
hand we know nothing of the nervous system of several principal groups of the Medusa?
(for example, of the two orders of Stauromedusse and Perornedusse) ; on the other hand,
in the remaining orders, the nervous system has not been examined on important
parts of the body, on which from their greater mobility and great sensibdity it is
probably very highly developed, pre-eminently on the oesophagus and the oral organs.
As far as we can judge at present the nervous system of the two sections presents
essential differences, as it appears more strongly centralised in the Craspedotae, more
diffuse in the Acraspedae.

§ 73. Nervous system of the Craspedotae. In all Craspedotaa, of which the nervous
system has been minutely examined up to this time (and among these we find some
belonging to all the four orders), its important centre represents a double marginal
nerve ring, lying on the proper umbrella margin immediately outside the insertion of
the velum. It is covered externally by a ciliated sense epithelium, consisting of small
flagellate cells, and is divided by the supporting plate of the velum insertion into two
separate rings, an exumbral and a subumbral ring. The dorsal or exumbral nerve ring
(PI. IX. fig. 7, re' ; PI. XII. fig. 12, re') is the so-called upper ring (the outer or lower
in the normal position of the velum), and seems to be pre-eminently the central organ of
sense ; it contains smaller and scantier ganglion cells, also finer fibrillar, and specially
provides for the different organs of sense of the umbrella margin (namely, the auditory
clubs and the tentacles). The ventral or subumbral nerve ring (PI. IX. fig. 7, re" ; PI.
XII. fig. 12, re ) is the so-caUed " lower" ring (the inner or upper ring in the normal
position of the velum), and appears to be pre-eminently the motor central organ ; it
contains larger and more numerous ganglion cells, as well as several fibrdke, and provides
specially for the muscular system of the velum and the suburnbrella. The two nerve
rings are immediately connected by numerous fine filaments, which pierce the separating
fulcra! lamella of the insertion of the velum, and give out numerous filaments which
extend like a plexus and are in connection with many peripheric ganglion cells. In
many Craspedotas the nerve ring shows sbght swellings, which are perhaps radial
ganglia at the points of insertion of the tentacles (especially at the four perradial
and four interradial).

§ 74. Nervous system of the Acraspedae. The structure of the nervous system in

REPORT ON THE DEEP-SEA MEDUSAE. xlvii

the first two orders of the four orders of this section, the Stauromedusse and Pero-
Medusae, is next to unknown ; the two other orders seem to comport themselves in some-
what different ways. The Cubomedusse (Charybdeidse""aiid Chirodropidae, PL XXVI. figs.
25, 26) are distinguished by a strong simple subumbral nerve ring which runs above the
umbrella margin at a considerable distance from it. It lies embedded in a groove of the
subumbrella, whose muscular plate is interrupted by it, and consists of a clear axial
cord, two turbid cords of fibrillae (an upper and an under) lying on the former and a
peculiar overlying nerve epithelium. The nerve ring is swollen at eight places into eight
ganglia. The four perradial ganglia are larger and lie higher at the bases of the four
highly developed sense clubs ; they send out sensible nerves to the sense clubs and motor
nerves to the muscular plate of the subumbrella. The four interradial ganglia he deeper at
the basis of the four strong tentacle pedalia, and send out both sensible and motor nerves
to the umbrella margin and the tentacles. Wide-spread plexus of fibrillae, in which numer-
ous multipolar and fusiform ganglion cells are situated, he in the subumbrella and the
velarium, and are connected with the nerve ring and its eight ganglia. The nervous system
of the allied Peromedusae, where we may expect to find the nerve ring in the depth of
the exumbral coronal furrow or at the coronal muscle, is probably of the same nature as
that of the Cubomedusae. On the other hand, the" nervous system of the Discomedusae,
which has been often examined, varies in its nature in so far that the nerve ring retreats
whilst the principal sense clubs (four perradial and four interradial) appear in the fore-
ground as eight separate marginal nerve centres. Each of these eight sense clubs or
marginal bodies in the Discomedusae contains in itself the organs of sense described
below, and its base encloses an independent nerve centre, which here, and as in the
Cubomedusaa, may be termed the principal ganglion. This consists of a thick pad of
nerve fibrillae and ganglion cells, which are in iumiediate connection both with the under-
lying tactile cells of the ectodermal sense epithelium and with the remaining organs of
sense of the rhopalium. Other filaments connect it with the nervous plexus of the
subumbrella, which extends between the ectodermal epithelium and the muscular plate
of the latter, and contains large motor ganglion cells. The bundles of fibrillar which
form immediate connection between the eight principal ganglia of the rhopalia and
correspond to the strong nerve ring of the Cubomedusae may be looked for in the
Discomedusae in the bottom of the umbrella cavity.

§ 75. Organs of sense (" sensillae "). All Medusae possess organs of sense on the um-
brella margin. The umbrella margin itself is covered for the most part with sense epithe-
lium ; it is the mother-ground and place of origin of different sensillae. These appear in the
simplest (and almost universally spread) form as tentacles, which are plainly homologous
to the margin tentacles of the polyps from which they have originated phylogenetically.
The sensillae are represented only by tentacles in few groups of Medusas. In most groups
besides tentacles we find differentiated organs of sense on the umbrella margin, which

xlviii THE VOYAGE OF H.M.S. CHALLENGER.

have proceeded partly from undeveloped tentacles, and have partly arisen independently
of these. As far as we are able at present to decipher the difficult physiological signifi-
cance of the different organs of sense, we can distinguish four categories of organs of sense in
the Medusae, according to the specific energy of each, viz. : — 1. Organs of touch, mechani-
cal tools for the perception of touch and pressure ; such, above all, are the tentacles with
manifold ectodermal cell formations, which appear specially adapted for the perception of
mechanical stimulation — viz., tactile bristles, tactile combs, &c. Besides the tentacles,
special tactile organs are often found at many places (namely, at the umbrella margin and
the margin of the mouth). 2. Organs of smell or organs of taste, chemical sense tools for
the perception of the mixture or rarefaction of the sea-water ; these are probably always
present (perhaps hidden under the tentacles already mentioned) ; the clavellae of the
Craspedotae, and the funnel-shaped depressions on the rhopalar protective scale of the
Acraspedae, may be perhaps regarded as special olfactory organs. 3. Organs of vision.
Ocelli or pigment eyes, with or without a lens principally and widely extended on the
umbrella margin and the basis of the tentacles ; sometimes adapted for thermatic percep-
tions, sometimes for optical (eyes for warmth, eyes for light). 4. Organs for hearing,
appearing in several different forms on the umbrella margin, among which we can
distinguish two originally different types, velar auditory vesicles with ectodermal otolites
and tentacular auditory clubs with endodermal otolites. All the four kinds of organs of
sense may be found united in one and the same " sense tentacle," as is the case with the
" sense clubs " or " rhopalia " of many Acraspedae. As the umbrella margin is the site of
the greater number of different sensillae, in the Medusa? they were usually given the
indifferent name of marginal bodies (" corpuscula marginalia"), which only indicated
their situation.

§ 76. Tentacles (t). The tentacles or feeling filaments are by far the most important
organs of the umbrella margin of the Medusae, as they not only represent the oldest and
simplest organs of sense of this class of urticating animals, but at the same time of their
limbs. The tentacles are originally placed on the umbrella margin (" marginal
filaments "), and are used as feelers as well as sense organs, also as weapons for attack
and defence, as sucking-cups for adhesion by suction (Pectyllidae, Pis. III.-VIIL), as
steering organs for swimming, or as manducatory organs for leading the nourishment
seized on to the mouth (" filaments of prehension "). Only the small group of the
Amaltheida? among the Craspedotae, and the large group of the Ehizostomae among the
Acraspedae are distinguished by complete absence of tentacles ; in them the tentacles
have undergone retroorade formation and become lost. As the tentacle corona of the
Medusae corresponds to that of their ancestors, the polyps, the conditions of formation
and structure are in general the same in both classes. The tentacles in most Medusas are
placed in a circle on the umbrella margin, just as they are placed on the corresponding
peristomial margin or calyx margin in most polyps. In the Medusa? they usually form a

REPORT ON THE DEEP-SEA MEDUSAE. xlix

single row, and are regularly distributed in it according to number and arrangement
(§ 77). More rarely two or more rows of tentacles are placed on tbe umbrella margin, tbe
one above the other, and are then usually compacted in larger numbers (Pectyllis, Pis.
III., IV. ; Pedis, Pis. V., VI.). The tentacles sometimes appeared grouped together in
bushes or bundles on the umbrella margin, as in the Lizusidae and Hippocrenidae (System,
taf. v., vi.), and in Pectanthis (Pis. VII., VIII.) among the Craspedotae ; and in the
Lucernaridae (Pis. XVI., XVII.) and the Chirodropidae (System, taf. xxvi.) among the
Acraspedae. Deviation from the original marginal insertion sometimes takes place, as the
tentacles either migrate outwardly on the dorsal surface of the umbrella or inwardly on
the ventral surface. Exumbral insertion, on the dorsal surface is found in many
Trachomedusae and most Narcomedusse (Pis. IX. -XIV.) ; there the tentacles may be
placed far up on the exumbrella, but usually denote their original connection with the
umbrella margin by the urticating streaks or umbrella clasps already mentioned
(" peronia," § 68 ; Pis. IX., XIII. , XIV., en). In the Aurelidse the tentacles are also
inserted dorsally (System, taf. xxxii. fig. 8). The Sthenonidae and Cyaneidae are distin-
guished by subumbral insertion of the tentacles ; in the latter they are scattered over
nearly the whole subumbrella (Drymonema, Pis. XXX., XXXL).

§,77. Number and position of the tentacles. Although the tentacles of the Medusae
present the most varied conditions both as to number and position, still by critical
comparison we are able to recognise the existence of certain simple primary and original
conditions, from which all the others may be secondarily derived. We may conclude from
this that most probably four perradial tentacles (at the distal end of the four radial
canals) represent the primitive formation for the Craspedotas, but eight principal tentacles
(four perradial and four interradial) for the Acraspedse. In the section of the Craspedotae
tetranemal forms (with four perradial tentacles) are found in all four orders ; Codonium,
Cytceis, &c, among the Anthomedusae, Tetranema, Eucopium, &c, among the Lepto-
medus£e, Petasus among the Trachomedusae, Cunantha among the Narcomedusae (comp.
System, p. 359); the two latter may, however, be regarded as already octonemal as in
them four interradial cordyli alternate with the four perradial tentacles, the cordyli
themselves being merely modified acoustic tentacles (§ 84, comp. PI. IX.). This is also
the case in Pericolpa, one of the oldest and simplest forms among the Acraspedae. The
inverted condition is shown in CJmrybdea (PI. XXVI. where the four sense clubs are
placed perradially, but the four tentacles mterradially. Both the Charybdeidas (Cubo-
medusae) and the Pericolpidae (Peromedusae) are derivable from Tessera, the oldest
and simplest form among the Stauromedusae, which may at the same time be considered
the hypothetic ancestral form of all Acraspedae. This has already eight principal
tentacles (four perradial and four interradial). In Pericolpa only the four interridal
tentacles are transformed into sense clubs, in Cliarybdea the four perradial, in Ephyra
(the ancestral form of the Discomedusae) all the eight principal tentacles. In the latter,

(ZOOL. CHALL. EXP. — PART XII. — 1881.) M g

1 THE VOYAGE OF H.M.S. CHALLENGER.

as in the closely allied Nauphanta (Pis. XXVII., XXVIII.), eight adradial tentacles (fig. B,
to) are developed between the eight principal. In the majority of the Medusae the
number of the tentacles increases with age, as new tentacles are formed later between the
original four or eight. This increase takes place according to fixed laws, which vary
in the different principal groups. The ontogenetic series in the appearance of the
different orders of tentacles, allows us to conclude that there is a corresponding phylo-
genetic progression. In contrast to the eight principal tentacles (four perradial and four
interradial) all the others which appear between them later may be termed succursal.
This distinction is important because the eight principal tentacles give rise to numerous
transformations and progressive formations. By retrograde formation of two opposite
perradial tentacles, dissonemal Medusa?, e.g., Thamnostylus dinema (PI. I.) often arise from
tetranemal. Such Medusa? are found among many groups of the Craspedota?, but not
among the Acraspeda?. The remains of the retrograded tentacles usually persist as
bulbs of the umbrella margin between the two opposite permanent tentacles. On the
other hand, we very rarely find only a single tentacle in the developed Medusa?, the
three others having undergone retrograde formation ; the Euphysida?, a small sub-family
of the Codonida? (System, taf. ii.) are mononemal Medusa?. The Amaltheida? among the
Craspedota?, and the Ehizostoma? among the Acraspeda?, are distinguished by complete
loss of all the tentacles.

§ 78. Form and structure of the tentacles. In most Medusa? the tentacles are long,
cylindrical filaments, more rarely flattened like a ribbon. They are usually thicker
at the base, but pointed conically towards the end, more rarely swollen like a club.
They are almost always simple and unbranched ; only a single family of the Craspedota?,
the Cladonemida?, are distinguished by branched or composite tentacles (System, taf. vii.) ;
these are sometimes branched dichotomously, sometimes beset with " secondary filaments "
(semi-pinnated) as in the Siphonophora and Ctenophora. The structure of the tentacles
is essentially the same in all Medusa?. They are composed of the same four essential layers
of tissue (or " secondary germinal layers") as the umbrella itself, namely: — (l) the outer
epithehum of the endoderm ; (2) the muscular plate underlying it, formed of longitudinal
fibres ; (3) the structureless, elastic supporting plate ; (4) the inner cellular axis of the
endoderm. We distinguish two principal forms, sobd tentacles and hollow tentacles ;
both often appear in closely albed Medusa?, sometimes beside one another in one and the
same species (Geryonida?). They are principally to be distinguished by the comportment
of the endodermal axis. The solid tentacles are usually stiffer and shorter, less extensible
and flexible ; they are chiefly found in the Trachylina? (Trachomedusa? and Narcomedusa?)
and also in the oldest forms of the Acraspeda? (Stauromedusa? and Cannostoma?). Their
cylindrical endodermal axis usually consists of a single row of discoid chordal cells, lying
the one above the other like the coins in a rouleau of sovereigns (PI. VI. fig. 17 ; PI.
XIII. figs. 5, 6, &c. ; PI. XV.). They are more rarely arranged in several layers (PI. IV.

REPORT ON THE DEEP-SEA MEDUSAE. li

figs. 5-8). The hollow tentacles are generally more flexible and movable, longer and
much more extensible, they are chiefly found in the Leptolinae (Antkomedusae and
Leptomedusse), and also in the majority of the Acraspedse. They contain a canal, which
represents a peripheric process of the gastrovascular system, and is lined by a single
layer of endodermal flagellate ceUs (PI. VII. fig. 4 ; PI. XVII. figs. 15, 16 ; PI. XXI. fig.
21, &c). In the two forms of tentacles, both solid and hollow, the endodermal axis is
covered by a structureless elastic supporting plate, which separates it from the overlying
muscular plate, and which at the same time acts as antagonist, or elastic extensor
against the contractions of the latter. The muscular plate consists of longitudinal
muscular fibrillar, which are usually still connected with the overlying epithebal muscular
cells of the ectoderm. The latter, moreover, contains thread cells and feeling cells very
variously arranged, and often also glandular cells and ciliated cells.

§ 79. Organs of feeling (tactile organs, " organa palpantia "). As sensibility to
variations of temperature, and reaction against touch and pressure is wide spread among
the Medusas, tactile cells (" cellulse palpantes ") must necessarily be generally pre-
sent. All indifferent sense cells, all ectodermal cells with hair-shaped processes may
probably be considered as such. These tactile hairs may be either flexible and movable
(" flagellum ") or stiff and immovable (tactile bristle, " palpellum "). Whether all ecto-
dermal flagellate cells are to be regarded as tactile cells is still doubtful, but this view
probably holds good for the flagellate cells composing the sense-epithelium above the
nerve ring of the umbrella margin and the "marginal corpuscules," and also for the
flagellate cells, which in many Medusas form part of the outer epithebum of the ten-
tacles (sometimes arranged in longitudinal streaks, rings or spirals along the sides
of the tentacles, sometimes as a connected covering of the ends of the tentacles).
We appear more justified in considering these ectodermal flagellate cells as tactile
cells, when we perceive that their bases are directly connected with nerve fibrillar
This is also the case with the " cells with tactile bristles " of the ectoderm, which bear a
stiff, often long, and far projecting tactile hah', a tactile bristle or palpellum. Such cells
with tactile bristles are usually widespread in the ectoderm both on the exumbral dorsal,
and on the subumbral ventral surface, chiefly, however, on the most sensitive parts, on
the umbrella margin and the tentacles, and also on the oral margin and the oral arms.
According to this view, the whole urticating cells in the first place, and in the second
place the " tactile cells" in the more limited sense, or the palpocells (without nematocysts)
belong to this category. The " urticating bristle " (" cnidocilium ") of the urticating cells,
like the feeling bristle (" palpocilium ") of the actual " feeling cells," is a direct process of
the protoplasm of the cell, projecting externally freely into the water and as in both cases
the latter is connected at the base of the cell with nerve fibrillar, in both cases the stimu-
lation received by the palpellum can also be communicated by the nerves to other parts
(muscles, &c). The urticating cells (with cnidocilia and nematocysts) and the feeling

lii THE VOYAGE OF H.M.S. CHALLENGER.

cells (with palpocilia but without nematocysts) are therefore to be regarded as two differ-
ent modifications of cells with tactile bristles (with palpella). The distribution of these
tactile cells on the sensitive organs of the umbrella and its appendages varies re-
markably. For example, in the Trachomedusae and Narcomedusae special " tactile
combs " are found on the umbrella mai-gin, or comb-shaped rows of tactile bristles, tactile
rings on the tentacles, and so forth (System, taf. xvii. figs. 9, 10, &c).

§ 80. Organs of smell (" organa olfactoria"). The peculiar chemical sensilke of the
Medusae, which might perhaps be equally or more truly termed organs of taste
(" organa gustatoria ") belong to this category. It is easily seen from physiological obser-
vations and experiments that the Medusae are very sensitive to change of composition of
the salt water, and even to slight rarefaction of it, so that, for example, they sink below
as soon as it begins to rain. The organs of chemical perception of sense are not yet
known with any certainty, and are probably usually represented by sense cells of the
umbrella margin, of the tentacles or of the margin of the mouth. Special organs, which
give the impression of sensillae from their situation and composition, may probably also lay
claim to this function, such as the marginal clavellae among the Craspedotae and the
rhopalar olfactory depressions among the Acraspedae. The " olfactory clubs " or mar-
ginal clubs (" clavelli marginales ") are only found in the section of the Craspedotae and
there chiefly in the order of the Leptomedusae. In my System, 1879 (pp. 118, 123, 143,
taf. viii. figs. 7, 12 ; taf. ix. figs. 3, 8) these clavellae were termed marginal clubs ("cor-
dyli marginales") as they are found in those Leptomedusae, in which the auditory
vesicles, which they therefore perhaps represent, are wanting (Thaumantidae, Cannotidae).
They usually (or always ?) want the characteristic " auditory hairs," which are a distin-
guishing feature of the acoustic organs. The pyriform or club-shaped clavellus (PL II.
figs. 3, 4, 8), sits with a thin stalk on the umbrella margin, and is, therefore, not to be
confounded with the conical disposition of the young tentacles. It contains a caecal, very
narrow " canalis clavellaris," which runs out from the coronal canal and is lined with
high cylindrical epithelium (PI. II. fig. 8, y). The latter is separated by a thin fulcral
plate (z) from the flat epitheluim of the ectoderm (q). The clavellae are found in many
Thaumantidae and Cannotidae on the umbrella margin, scattered in large numbers (often
several hundreds), between the tentacles, and may, therefore, be regarded as sensillae.
This is also applicable to the olfactory depressions (" fossulaa olfactorise, oz) of the
Acraspedae. These appear in the Discomedusa? as small, caecal funnel-shaped depressions
in the dorsal surface of the rhopalar protective scales (or " funnel plates "), and are
Uned with a many folded sense-epithelium, furnished with long flagella (conip. under the
rhopalia).

§ 81. Organs of vision (" organa optica "). Physiological experiments prove easily and
with certainty that all Medusae are more or less sensible to the influence of light and
warmth. From analogy with other animals we are justified in the conclusion that the

REPORT ON THE DEEP-SEA MEDUSAE. liii

simplest organs of this sensation are the pigment spots ("ocelli"), chiefly those placed on
the umbrella margin. They consist partly of pigment cells, partly of optical sense
cells or root cells, which belong to the sense epithelium of the dorsal nerve ring. Whilst
these ocelK are originally simple centres for the perception of heat, they are developed
later on into true light eyes. As experiments showed, it is principally the swollen bases
of the tentacles which bear such pigment eyes, and that chiefly in the order of the Antho-
medusae and in those Leptomedusae which have no marginal vesicles (" ocellatae "). Such
ocelli are more rarely found in the Trachomedusae, Narcomedusas, and Stauromedusse.
On the other hand they are widely spread among the three higher orders of the
Acraspedae and usually found at the base of the sense clubs or rhopalia described
below. In many Acraspedae and a few Craspedotaa (Anthomedusae) a lens is also found
in the pigment body of the eye and in the Cubomedusae a crystalline lens or retina is
developed between the lens and the pigment cup. We also find " composite eyes " in
the Cubomedusae, as, e.g., in Charybdea, where each sense club bears two large unpaired
and two small paired eyes. Moreover, Medusae perfectly devoid of colour, which have
neither marginal ocelb nor other pigment spots, are sensible to light ; in this case it is
probably the sense epithellium of the umbrella margin which discharges this function.
We therefore find in the class of the Medusae a long series of different phylogenetic
stages of development of optical apparatus, from the simplest beginning up to very
composite eyes.

§ 82. Organs of hearing (" organa acustica "). In the majority of Medusae we find
organs of sense on the umbrella margin, which must be indubitably regarded as organs
of sense as they possess both otobtes (" otobthi ") and auditor)7 cells (" otocellae ")
bearing bristles. In the minority of the Medusae, in which the otolites are absent,
it is possible (or rather probable) that a lower degree of acoustic functions are
exercised by part of the cells bearing tactile bristles (" palpocellae ") already described.
As, on the one hand, we know of no definite morphological distinction between such
tactile cells, bearing bristles and auditory cells which also bear bristles, and as on the
other hand, the latter must be regarded as merely special modifications of the former, it
is possible that many apparently indifferent tactile cells are sensible not only to fluctua-
tions of pressure, but also to vibrations of sound. Considering, however, the immense use
which the capacity of hearing must be to the free swimming Medusae (e.g., the perception
of the noise of the tempestuous breakers on nearing the coast), it is most probable that
a lower or higher degree of sensibility to sound is generally spread in this class. In this
case we must consider the cells with tactile bristles, which are found in the Medusas
devoid of otolites, in the whole of the Anthomedusae, Stauromedusas, and also in the Ocellatae
(Thaumantidae and Cannotidae) must among the Leptomedusas be regarded as " lower
acoustic organs." All other Medusae, on the contrary, possess " higher acoustic organs "
or true " organs of hearing," consisting of auditory cells and otolites ; these are found in

liv

THE VOYAGE OF H.M.S. CHALLENGER.

all the Acraspedse (with the single exception of the Stauromedusse) and in the majority
of the Craspedotse, in all Trachomedusse and Narcomedusae, and also in the Vesiculatse
(Eucopida? and ^Equoridaa) among the Leptomedusse. In the Vesiculatse, however, the
organs of hearing have quite a different structure and a different origin from those in
the other groups mentioned, the Vesiculatee have velar auditory vesicles with ectodermal
otolites, whilst all the others have tentacular auditory clubs with endodermal otolites.
These two types in the structure of the auditory organs differ so entirely that they
require a separate description.

§ 83. Velar auditory vesicles (with ectodermal otolites, " vesiculae velares," also
termed " marginal vesicles," " vesiculae marginales," ov). This peculiar form of the audi-

Fig. E. Eucopc campanulata (Leptomedusre, Eueopidse).

Subumbral aspect, («) Quadrate oral opening, (s) Ovaries, (cp) Perradial canals, (cc) Coronal canals.
(v) Velum, (ov) Velar marginal vesicles (adradial). (7s) Umbrella cavity, (re) Nerve ring, (oc)
Ocelli at the swollen bases of the tentacles, (ug) Gelatinous substance of the umbrella. {I) Tentacles.

tory organs is found exclusively in the order of the Leptomedusae, and there only in a
suborder which we name Vesiculates for this reason (the two families, rich in forms, of
the Eucopidse and iEquoridae, System, taf. x.-xv. pp. 116, 165, 210). The other sub-
order of the Leptomedusse, the Ocellatse (the two families of the Thaunianthidae and
Cannotidse) have neither otolites nor auditory vesicles, but perhaps auditory cells on the
subumbral side of the velum. This exception is supported by the fact, that in the simplest
cases in the Vesiculatse (Mitrocoma) no closed auditory vesicles are present, but only open
depressions in the insertion of the velum, in whose ectoderm auditory cells and otolite

REPORT ON THE DEEP-SEA MEDUSAE. lv

cells are developed. These auditory depressions (" fossulae velares ") are probably found
in other Leptomedusae (Phialis, Tiaropsis, Mitrocomella, &c), besides Mitrocoma
(System, taf. x.). They form small depressions in the subumbral or ventral side of the
velum (which is commonly termed the " lower " side but which is the " upper " in the
normal position of the horizontally stretched velum). One portion' of the subumbral
ectoderm cells, which line these depressions, and which are connected with the contiguous
ventral nerve ring, forms a calcareous otolite in their interior, another portion of it bears
an auditory bristle. As these " auditory depressions " become deeper, enter the dorsal
side of the velum vaulted like an arch, and finally become entirely loosed from the ventral
side, they are transformed into auditory vesicles ("vesicuke velares"). These project
more or less as conical or roundish vesicles on the dorsal side of the velum, near its inser-
tion on the umbrella margin, become innervated from the subumbral nerve ring, covered
externally by the dorsal ectodermal epithelium of the velum and contain a hollow space
filled with otolymph (originally sea water) ; this space is lined with an acoustic
epithelium, which originally belongs to the ventral ectodermal epithelium of the velum
and consists partly of auditory cells bearing bristles and partly of otolite cells. The
auditory hairs of the former surround or He upon the latter. In the most simple cases,
each velar marginal vesicle only contains a single otobte, but in others often a large number
of them. The inner (subumbral) sense epithelium and the outer (exumbral) covering
epithelium, are separated by a structureless lamella which belongs to the supporting
plate of the velum. The number and distribution of these velar auditory vesicles of the
Leptomedusae varies largely ; however, there are always originally eight adradial audi-
tory vesicles, which lie exactly in the middle between the four perradial and the four
interradial tentacles (fig. E). We never find fewer than eight. In most Leptomedusae
their number is considerably increased, often to several hundreds ; we may therefore dis-
tinguish two groups of the Vesiculatae, the Octotessae, and Polyotessae, the former having
invariable eight velar auditory vesicles, the latter having invariably more than eight
(System, p. 117, taf. x.-xv.).

§ 84. Auditory clubs or " cordyli " (ok), acoustic tentacles with endodermal otolites.
This form of the organs of hearing is by far the most common among the Medusae, and is
found in the majority of the class, in the Trachomedusae and Narcoruedusae among the
Craspedotaa, and also among all the Acraspedae, with the single exception of the Stauro-
medusae. The auditory clubs of all these Medusae are modified, small acoustic tentacles,
containing endodermal otolites, and differing therefore entirely in origin and composition
from the velar auditory vesicles of the Leptomedusaa (with ectodermal otolites). In the
two sections, the auditory clubs have originated, independently of one another, from the
tentacles in an analogous manner, from the solid tentacles in the Craspedotae, from the
hollow tentacles in the Acraspedae. In the former, it is the chordal cells of the solid
endoderm axis, which produce the otolites, in the latter it is the endodermal cells form-

lvi THE VOYAGE OF H.M.S. CHALLENGER.

ing the epithelium at the distal end of the tentacle canal. The auditory clubs of the
Acraspedse are, moreover, combined in a peculiar way, with other organs of sense (ocelli,
olfactory depressions, tactile plates) and with their surroundings, compose the typical
sense clubs or rhopalia, which we shall afterwards consider separately. On the other
hand the analogous auditory clubs of the Craspedotse (of the Trachomedusae and Narco-
medusae) which often closely resemble the others, appear to be more simple formations
(Pis. III. -XI V., oh). They resemble perfectly simple, sobd tentacles, whose axis consists
of a row of a few endoderm cells (usually two to four, more rarely five to ten or more),
processes of the inner epithebum of the coronal canal. Either only the last of these or
several of them (two to four, rarely more) produce in their interior a calcareous concre-
ment, which functions as an otolite. The Trachomedusae (and a small part of the
Narcomedusae, the Solmaridse) have only a single round otolite in each auditory club ;
it is in concentric layers, usually spheroidal, more rarely elliptical, and often coloured
red or yellow. Most Narcomedusae (all, indeed, except the Solmaridae) possess crystalline
otontes of prismatic form (usually several in each auditory club). The acoustic ectoderm
epithelium of the auditory clubs is separated from the sobd endoderm axis by a thin
supporting plate, and bears long stiff auditory hairs ; so is the ectoderm of the " auditory
pad " or " auditory papilla," which in many Narcomedusae arises at the base of the
auditory club by a swelling of the dorsal nerve ring ; the latter always supphes the
cordylus. In part of the Narcomedusae (the Cunanthidae and Peganthidae) peculiar, firm
urticating streaks are found at the bases of the cordyli which rise from these centripetally
into the exumbrella, and are covered with ciliated sense epithelium (auditory clasps,
" otoporpae," PI. IX. fig. 8, oo ; PI. XL fig. 4, 06). Four interradial auditory clubs seem
usually present originally ; the number often increases largely later on, and may amount
to more than a thousand (e.g., PegantJm magnified, System, p. 333).

§ 85. Cordylar auditory vesicles ("vesiculae cordylares"). Whilst in all Narcomedusae,
and also in the lower and older groups of the Trachomedusae (Petasidee, Pectyllidae, Pis.
III. -VIII., Aglauridae), the auditory clubs stand freely on the umbrella margin, in
some of the younger and higher groups of the Trachomedusae, this is rarely the case,
and then only in the young stage. The originally free auditory clubs become en-
closed in special " auditory vesicles." In the Marmanemidae (System, taf. xvii.) this
is caused by the ectodermal epithelium of the dorsal nerve ring rising like a wall, in the
form of a circular fold, at the base of the free cordylus ; its margins grow together above
the depression formed, and so transform it into a closed vesicle ; the auditory hairs are
stretched like harp strings between the inner wall of the vesicle and the upper surface of
the cordylus enclosed. Whilst these " auditory vesicles " of the Marmanemidae lie freely
on the umbrella margin, the similarly constructed auditory vesicles of the Geryonidae lie
deeply inserted in the gelatinous body of the umbrella margin. These cordylar auditory
vesicles of the Marmanemidae and Geryonidae therefore differ entirely both in origin and

REPORT ON THE DEEP-SEA MEDUSAE. lvii

in finer structure from the " velar auditory vesicles " of the Lcptomedusse (§ 83) with
which they were formerly usually confounded.

§ 86. Sense clubs (" rhopalia," or). We designate by this name the peculiar, " composite
organs of sense," or " marginal bodies " of the Acraspedse, which are always universal in
this section, and only wanting in the lowest and oldest Acraspedse, the Stauromedusse.
In place of rhopalia the Stauromcdusee have simple tentacles, and it seems undeniable
from their whole structure, situation, and distribution that the rhopalia of the Acraspedse
are modified tentacles furnished with several different organs of sense. If we assume
that the Tessera, the simplest and oldest among the known Acraspedse, is the common
ancestral form of this section (or at least does not differ essentially from the hypothetic
ancestral form), the characteristic position of the sense clubs in three higher orders of the
Acraspedse is explained as follows : — Of the eight principal tentacles of Tessera the four
interradial are transformed into rhopalia in the Peromedusse (Pis. XVIII. , XIX.), and the
four perradial in the Cubomedusse (PI. XXVL), whilst the four tentacles alternating with
them remain unaltered. In the Discomedusaa, on the other hand, all the eight tentacles of
Tessera have become sense clubs ; in fact the majority of the Ephyroniae have four per-
radial and four interradial sense clubs (Pis. XXVII.-XXXII.), and it is only in a few
genera that their number increases secondarily from twelve to sixteen, rarely from twenty-
four to thirty-two (System, pp. 364, 401, 427, 457). As the sense clubs of the Acraspedse
in this section have originated independently, and as even the four perradial rhopalia of
the Cubomedusse have been formed from " acoustic tentacles " independent of the four
interradial sense clubs of the Peromedusse, the former present no homology with the
similar auditory clubs or cordyli of the Craspedotae, but only a close analogy ; they are
distinguished from the latter by their more composite structure, and also by their pro-
tected position in special rhopalar niches (hence Steganophthahnse). The rhopalar niches
(" antra rhopalaria," PL XXX. figs. 2-4, on) are ectodermal cavities, which lie in most
Acraspedse on the umbrella margin, but which sometimes change their marginal position
later on and migrate either on to the dorsal surface of the exumbrella (Cubomedusse, PI.
XXVL) or on to the ventral surface of the subumbrella (Drymonema, Pis. XXX., XXXI. ).
The sense niches or sense sinuses are enclosed on both sides, usually on their ventral or
axial surface, by the paired " sense-folds," the axially projecting medial margins of a pair
of sense lobes of the umbrella margin (rhopalar lobes) ; these " plicse rhopalares " (of) are
sometimes fused into a plate. On the other hand, the unpaired sense scale or protecting
scale (" squama rhopalaris," os), originating from the marginal bit of the exumbrella, which
originally formed a connecting bridge between the two sense folds, projects on the dorsal
or abaxial side of the rhopalar niche as a protecting roof. In the convex dorsal surface
of the protective scale, there is usually a csecal funnel-shaped olfactory depression
(" fossula olfactoria," os) whose folded sense-epithelium is furnished with special flagellate
cells (olfactory cells). The true sense club, which lies hidden in the niche, corresponds

(ZOOL. CHALL. EXP. — PABT XII. — 1881.) M k

lviii THE VOYAGE OF H.M.S. CHALLENGER.

to a short, club-shaped hollow tentacle, whose " sense canal " ends in an otolite sac or
" crystal sac " (PL XXX. figs. 4-7). The latter consists of a considerable, spheroidal or
oval accumulation of crystalline concrements, which have been formed in the endoderm
cells of the tentacle canal ; it is enclosed in a fulcral sheath covered externally by the
ectodermal epithelium, beset with long, stiff, auditory hairs. A peculiar tactile plate (?)
whose rod-shaped tactile cells bear long flagella, is usually found at the proximal base of
the auditory club on the axial ventral side, whilst on the abaxial dorsal side there is a
visible pigment pad which is considered as an eye, and sometimes encloses a lens and
sometimes not. These eyes appear to attain their highest development in the Pero-
medusse and Cubomedusse, in them we often find several eyes in each rhopalium, in which
a crystalline lens and a retina with a large optic ganglion may be developed (System,
pp. 401, 427 ; taf. xxiii., xxv., &c.,).

§ 87. In all Medusse the muscular system is composed of two different principal
sections, a circular, and a longitudinal system of fibres. Both form a thorough
contrast, not only by their local distribution and by the direction of the course of their
fibres, but also by their histological nature ; the circular or transverse fibres are usually
clearly striated, whilst the radial or longitudinal fibres are fiat for the most part. By far
the largest and most important part of the two systems belong to the subumbrella which
functions chiefly as swimming organ. The muscular system of the umbrella margin
and the tentacles generally proceeds from the subumbrella. On the other hand, the
muscular system of the exumbrella, which is only very partially developed, is by no
means important. Both the transverse and the longitudinal fibres are exclusively pro-
ducts of the ectodermal epithelium, with which they are still most closely connected (comp.
above, §§ 51-53). Moreover, in some (perhaps all?) of the Medusae, weak (usually very
unimportant) muscles which originate from the endodermal epithelium of the gastro-
vascular system appear on certain parts of the body. Certain circular muscles of the
oesophagus and the muscles of the gastral filaments belong to these endodermal muscles,
which as yet have been but little recognised and investigated. Although the two sec-
tions of the class of Medusa? have originated independently of one another, the differen-
tiation of the muscular system show analogous conditions in both cases. In both the
circular and the radial system of fibres we can generally distinguish three sections, of
which the first occupies the central and proximal part, the second the middle part (the
true umbrella in the more restricted sense), and the third the marginal or distal part
(along with the marginal appendages).

REPORT ON THE DEEP-SEA MEDUSAE.
Survey of the two muscular systems of the sul mmbrella.

lix

I. System op Circular Muscles

II. SrSTEM OF

Radial Muscles

(composed of striated fibres, running transversely
or circularly).

(composed of smooth fibres, running longitudinally
or radially).

rl. " M. orbiculares," oral

rl.

" M. proboscidales," probos-

1.

Proximal part of

circular muscles (cir-

1.

Proximal part of

cis muscles (longitudinal

tho circular mus- ■

cular muscles of the

the radial mus- -

muscles of the oesophagus

cular system.

oesophagus and the
stomach).

cular system.

•

and of the gastral pe-
duncle.

2.

Middle part of the I 2. " M. coronares," coronal

2.

Middle part of / 2.

" M. codonoides," muscles

circular muscu- < muscles (coronal

the radial mus- <

of the bell (deltoid mus-

lar system. 1 pouches, &c).

cular system. 1

cles, &c).

3.

_. , , , « ,, /3. "M. velares," velar mus-
JJistal part of the ( . . . , ,
. 1 cles (circular muscles
circular muscu- < . ,, v , , . ,
, I or the velum and the
lar system. / , .

V velarium).

3.

3

Distal part of (

the radial mus- <J
cular system. {

" M. marginales," marginal
muscles (longitudinal
muscles of the tentacles

and the marginal lobes.)

§ 88. Circular muscles of the subumbrella (" myosystema circulare "). The circular
muscular system of the subumbrella is developed quite analogously in the two sections
of the Medusae class, and consists of transversely-striated muscles, running in horizontal
transverse planes (perpendicularly to the principal axis). This system is divided into
three different sections, — the proximal circular muscle of the oesophagus and the oral
organs (" musculus orbicularis "), middle coronal muscle (" musculus coronaris "), and the
distal circular muscle of the velum (" musculus velaris "). The circular oral muscle
(" musculus orbicularis," mo) forms the proximal part of the circular system, and is gene-
rally the weakest of its three sections and also the most irregularly developed in the
different groups. It only attains any considerable development in such Medusae as are
distinguished by a strong, movable oesophagus, or by large oral lobes or folded oral arms ;
for example, among the Craspedotse, on the folded oral lobes of many Anthomedusse and
Leptomedusae, on the raised lips, capable of extension into a large sucking-disk, of many
Trachomedusae (PI. III. fig. 2 ; PI. V. figs. 3, 4 ; PI. VII., fig. 3, am) ; and on the very
contractile and extensible oesophagus of many Narcomedusae (Pis. IX.-XIV.) ; among the
Acraspedse on the oral lobes of the Stauroniedusae and Cuboniedusae (Pis. XV., XVII. ,
XXVI.) ; on the buccal pouches of the Peromedusae (PI. XX. figs. 9-11), and on the oral
arms of many Discomedusas (Pis. XXX.-XXXIL). The second and middle section of the
circular muscular system, the large coronal muscle (" musculus coronaris," mc) is much
more important. It must be regarded originally in all Medusae as the most important
swimming muscle, and in most of them, it occupies the greater part of the subumbrella,
from the distal margin of the orbicular muscle (or in others of the bell-muscle) to the proxi-

lx

THE VOYAGE OF H.M.S. CHALLENGER.

mal margin of the velar muscle. Its parallel and thickly compacted circular fibres
sometimes run as uninterrupted rings on the whole subumbrella, or are sometimes divided
by four, eight, sixteen or more radial septa into an equal number of separate plates. The
coronal muscle is divided into four interradial plates in a large number of Craspedotse; into
eight adradial, e.g., in the Cubomedusse (PL XXVI.), into eight principal (four perradial
and four interradial) plates in Pericolpa (fig. F), into sixteen subradial in the Pectyllidse
(Pis. IV., VIII.) ; in Periphylki, on the other hand, eight of the coronal plates are principal,
and eight adradial (PI. XIX.). Very often (namely, in the larger Acraspedse) the sup-
porting plate rises below the coronal muscle in the form of simpler or more complex cir-
cular folds, so that the muscular system lying above it commands a more extensive

Fig. F. Pericolpa qitadrigata (Peromedusre, Pericolpidre).

Subunibral aspect, (oi) Sense clubs (interradial). (oa) Ampulla at their bases. (() Tentacles (perradial).
(bl) Lobe pouches, (bit) Horseshoe-shaped canals of the pouches. (M) Peronia between the two
limbs of the canal, (s) Genitalia, (mk) Root-muscles of the tentacles. (mdl) Perradial deltoid
muscles, (md?) Interradial deltoid muscles, (aft) Oral ribs, (ar) Oral grooves on their inner sur-
face, (ai) Oral tavniola. (tr) Tentacle roots, (be) Coronal pouches, (mc) Coronal muscles.

surface of insertions in a limited space (Pis. XIX. -XXII. , mc). In most Craspedotae, the
coronal muscle is comparatively broader, and covers the larger part of the subumbrella as
connected plates, whilst in most Acraspedse it is narrower and more limited to their peri-
phery ; it is often forcibly thrust back by the deltoid muscles, which extend themselves
at its expense (fig. F, md). The velar muscle (" musculus velaris ") forms the third
distal section of the subunibral circular muscular system. In the Craspedotas it repre-
sents the most important element of the velum (Pis. IV.-XIV.) and is separated by the
ventral nerve ring from the contiguous distal margin of the coronal muscle. It is repre-
sented in the Acraspedse, by the analogous (but not homologous) coronal muscle of the

REPORT ON THE DEEP-SEA MEDUSJ3. lxi

velarium ; it is most strongly developed in the Cubomedusse (PI. XXVI. , mv). Comp.

§§ 66-70.

§ 89. Eadial muscles of the subumbrella (" myosystema radiale "). Whilst the cir-
cular muscular system of the Medusae is almost invariable composed of transversely stri-
ated fibres, the radial muscular system in both sections of this class, is formed for the most
part of smooth, not transversely striated, muscular fibres ; these run in vertical meridian
planes, sometimes more radially, sometimes more longitudinally, parallel to the principal
axis. The radial muscular fibres are also divided into three sections analogous to those
of the circular muscular fibres ; the proximal longitudinal muscles of the oesophagus and
of the oral organs, and also of the gastral peduncle ("musculi proboscidales"), the middle
bell muscle ("musculus codonoides"), and the distal longitudinal muscles of the umbrella
margin ("musculi marginales"). The system of the proboscidal muscles ("musculi
proboscidales ") forms the proximal part of the radial system ; we may include in it, the
true longitudinal muscles of the oesophagus and of the different oral organs (oral lobes,
oral arms), and also the longitudinal muscles of the gelatinous gastral peduncle (e.g., in
the Octorchidae System, taf. xii., xiii. ; Geryonidae, System, taf. xviii.) ; in proportion as
the gastral peduncle is developed as a movable " proboscis," the four, six, or eight broad
longitudinal bands of muscles, which run in its upper surface between the ascending
radial canals, become more powerful. Of the longitudinal muscles of the oesophagus itself
the four perradial are usually most strongly developed. They run on the four folded
oral lobes or the large oral arms which have originated from them, on the abaxial side of
their midrib, and radiate, often in bunches, towards their frilled oral margin. In most
Medusae the second and middle section of the radial muscular system, whose separate
part we include under the name of bell muscles, is far more important. All the longitu-
dinal muscles of the subumbrella which lie between the base of the " proboscis " and the
umbrella margin belong to it. The bell muscle (like the coronal) not unfrequently (namely,
in a part of the Narcomedusae and Stauromedusae) represents a single, bell-shaped,
arched muscular plate, consisting of diverging radial fibres ; it is usually divided into a
number (four, eight, sixteen, or more) of separate longitudinal muscles. In most
Craspedotae four or eight such longitudinal muscles run on the subumbral surface of the
radial canals, but often also four or eight in the middle between these. We conse-
tpiently find usually four perradial longitudinal bands (w#>) and four interradial (mi)
between them, beside often eight adradial, rarely more. They are sometimes simple,
unpaired bands, which run exactly in the middle line of the radial canals (e.g., Pectyllidae,
Pis. III.— VIII.), sometimes paired bands, enclosing the two lateral margins of the
canals (e.g., Tiaridae, System, taf. iv. figs. 2, 3). The subumbral radial muscles are most
strongly developed in those Craspedotae which form circumoral buttresses (mesenteries or
mcsogonia). Many Tiaridae have four such perradial mesenteries, whilst the Pectyllidae
have eight principal mesenteries (PL IV. fig. 3, PI. VIII. fig. 9, wr). They lie as four or

lxii THE VOYAGE OF H.M.S. CHALLENGER.

eight thin, broad leaves in the principal radial planes, and run in the umbrella cavity,
stretching freely from the subumbrella to the oesophagus ; their longitudinal muscular fibres
(" musculi mesenteriales ") pass into the proboscis muscles, at the proximal margin of the
mesogonia. Among the Acraspedse the bell muscle of the Cubomedusse usually comports
itself like that of the Craspedotge (PI. XXVI. ; System, taf. xxv. xxvi.). The subumbrella
of the Tesseronia is on the whole almost cubical ; its coronal muscle consists of four
broad quadrangular(often almost quadrate or rectangular) muscular plates, which touch at
right angles in the interradii. They are divided by four narrow, interradial longitudinal
muscles, running along the four cathammal septa, but are halved in the middle by four
broader perradial longitudinal muscles ; the latter pass below upon four " frenula velarii "
(PI. XXVI. fig. 8, vf), above upon the mesenteries which run to the four corners of the
stomach (PI. XXVI. figs. 2, 3 ; System, taf. xxvi. figs. 2, 3, gm). In the remaining
Tesseronia (both in the Stauromedusse and the Peromedusge) the three strong, broad trian-
gular deltoid muscles (Pis. XV.-XXII. md) occupy the place of these narrow, band-shaped
longitudinal muscles. The deltoid muscles spring with a broad base at the proximal
margin of the coronal muscle (mc), and run with converging fibres towards the bottom
of the umbrella cavity (fig. F, md). The four interradial deltoid muscles (md) are usually
considerably stronger than the four perradial (md) ; the former are inserted at the four
interradial cathammal nodes (kn), and often pass out above them as " intergenital
muscles " (PI. XX. fig. 8, ms) ; the latter are inserted at the four perradial palatine nodes
(gk), and pass from them upon the mesenteries and the corners of the oesophagus. In the
Ephyronise or Discomedusse, those parts of the bell muscle appear much less important than
in the Tesseronia?, which is accounted for by the retrograde formation of the four perradial
pouches, and by the extension of the broad umbrella corona. The eight deltoid muscles
(and especially the four interradial) are pretty strongly developed only in a few Canno-
stomae (as in Atolla, PL XXIX. fig. 3, and Nauphanta, PI. XXVIII. fig. 12), whilst in
most Discomedusse they have undergone retrograde formation. We may therefore regard
the four strong pillar muscles of many Semostornse and KhizostomEe as developments of
the four perradial deltoid muscles ; they pass at their proximal end into the four perradial
proboscis muscles. The system of the marginal muscles (" musculi marginales ") forms
the third and distal section of the radial muscular system. Under the term " marginal
muscles " we include all the longitudinal or radial muscles which are developed on the
umbrella margin outside the circular coronal muscle. They are differentiated in many
varied ways. The most important are the muscles of the tentacles and of the marginal
lobes. The muscular fibres of the tentacles all run longitudinally in a great variety of
arrangements.

§ 90. Muscular system of the exumbrella. Development of the muscular system
appears entirely wanting on the upper or exurnbral surface of the umbrella, when con-
trasted with its powerful and universal development on the lower or suburnbral surface.

REPORT ON THE DEEP-SEA MEDUSAE. lxiii

Closer investigation, however, shows that muscles are also developed here in some places,
though but feebly, at least in some groups of Medusas (perhaps in all ?), and then into both
transverse and longitudinal cords of fibre. The principal exumbral circular muscles are
the feeble zonal muscles (" musculi zonares," mz) which are found in some Craspedotse
above the umbrella margin, in others in the circular stricture between the umbrella and
the apical process {e.g., Catablema, System, taf. iv. fig. 4). They are more strorjgly
developed in some Acraspedas as, for example, in the visible zonal muscle of the Pero-
medusse which divides the smooth umbrella cone from the pedal zone of the umbrella
corona and sends out zigzag processes between its pedalia (PI. XXIII. fig. 34, mz ; PI.
XXIV. fig. 2, mz). Exumbral radial muscles are found in some groups (especially in the
Trachomedusas and Narcomedusse), developed on the umbrella margin into peronial
muscles (" musculi peroniales," PI. XIII. fig. 7, ml; PI. XIV. fig. 12, ml). Other, but
feebler, longitudinal muscles appear in both sections here and there on the umbrella apex
and on other places of the convex outer umbrella surface. Among the Craspedotse, four
perradial and four interradial longitudinal muscles are found in the peripheric (and some-
times also in the central part) in some Anthoniedusse ; and among the Acraspedse, in the
Cubomedusse. The longitudinal muscular bands of the tseniola and the strong peduncle
muscles of the Stauromedusse, also belong to this system (Pis. XVI., XVII. figs. 13,
14, m).

§ 91. Umbrella cavity or swimming cavity (" antrum, caverna umbralis," h). The
umbrella cavity of the Medusas is as characteristic for this class of urticating animals as
the umbrella itself ; it is enclosed above by the lower concave surface of the umbrella
(" subumbrella"), whilst it opens freely below through the aperture of the umbrella cavity
(" apertura antri "). The subumbral umbrella cavity is more or less vaulted, according as
the umbrella is more umbrella shaped or more conical ; its vaulted roof, which is fined by
the ectoderm of the subumbrella, is, however, always flatter than the outer surface of the
umbrella which is covered by the ectoderm of the exumbrella, for the gelatinous wall of
peripheric umbrella corona is always thinner than the central umbrella cone. As by each
contraction of the swimming Medusas, the vaulting of the umbrella cavity becomes higher,
its opening narrower, and water is ejected through the opening, whilst by each dilatation
of the umbrella fresh water enters the flattened and widened umbrella cavity, the latter
may be regarded physiologically both as a " swimming cavity " and a " respiratory
cavity." The ectodermal epithelium of the subumbrella, which fines the umbrella cavity,
is probably adapted for respiratory functions. The opening of the umbrella cavity
(" apertura antri ") is simple and surrounded by the corona of lobes in the Acraspedse,
whilst in the Craspedotse it is narrowed by the velum, which projects freely inwards like
a diaphragm, from the umbrella margin. In some Craspedotse the velum is so broad,
that it is probably capable of completely closing, for a while, the opening of the umbrella
cavity, as in the Pectyllidas (Pis. III.-VIIL), The central axial space of the subumbrella

lxiv THE VOYAGE OF H.M.S. CHALLENGER.

is occupied more or less by the oesophagus and the different organs of buccal stomach and
also often by the genitalia.

§ 92. Niches of the umbrella cavity (" cavernulae subumbrales "). In many Medusae
special secondary spaces are developed on the subumbral wall of the umbrella cavity,
partly by the formation of folds or projections of the subumbrella, partly by the insertion
of single organs into pit-like depressions and partly by peculiar conditions of growth of
the umbrella margin and the " marginal bodies " lying on it. All these different second-
ary cavities of the umbrella cavity may be placed together as " niches of the umbrella
cavity or subumbral niches " (" cavernulae subumbrales "). In many Narcomedusae,
namely the Peganthidae, the cavity of the umbrella corona is divided into a peripheric
corona of separate "lobe cavities" (" cavernulae lobares"), which surround the central
umbrella cavity like the altar niches of a round temple (JPegantha pantheon, p. 37,
Pis. XL, XII.). In Pedis, eight adradial " oral funnels or inner buccal pouches " (" caver-
nulae buccales ") are invaginated from outside into the oesophagus (p. 15, Pis. IV., V. figs.
4, 5, io). In many Cubornedusse and Peromedusae, namely, the Periphyllidas, each tentacle
is surrounded at its base by a subumbral tentacle funnel (" cavernula tentacularis "), over
which the distal margin of the subumbral coronal muscle projects like a roof. In Periphjlla
(PI. XIX. fig. 6, PL XX. fig. 8) it is simple; in Periphema (PI. XXIV. fig. 1) it is divided
into secondary funnels by a number of small frenula. In many Cubomedusas, four perradial
triangular subumbral folds pass as " frenula velarii " from the base of the sense niche and
the vertical septum of the marginal pouches to the subumbral surface of the horizontal
inwardly projecting velarium (PI. XXVI. figs. 2, 3, 8, irf ), so that two small velar niches
(" cavernulae velares ") are inserted on each side of the velarium. In most Discomedusae
eight (more rarely sixteen) sense niches are formed on the umbrella margin for the recep-
tion of the sense clubs or rhopalia ("antra rhopalaria," comp. above, § 86). In some
species, e.g., in Drymonema, these stretch centripetally far into the subumbrella (Pis.
XXX., XXXL, cm).

§ 93. Coronal cavity of the umbrella and funnel cavity of the umbrella. In some
Craspedotae, or in many Acraspedae, four or eight vertical folds of the subumbrella, the
mesenteries (" mesenteria ") are developed in the bottom of the umbrella cavity at the
base of the oesophagus, and the upper part of the simple umbrella cavity is thus divided
into four or eight separate cavities, the umbrella funnels or funnel cavities (" infundi-
bula," i). We therefore term the lower, simple half of the umbrella cavity, which opens
freely below at the umbrella margin, the coronal cavity of the umbrella (" antrum
coronare "), and the upper quadrilocular or octolocular half as the funnel cavity of the
umbrella ("antrum infundibulare"); the former communicate with the latter by four
interradial or eight adradial funnel openings (" ostia infundibularia ").

§ 94. Funnel cavities and mesenteries (" infundibula and mesenteria "). The four or
eight funnel cavities or umbrella funnels (" infundibula," i), which compose the umbrella

REPORT ON THE DEEP-SEA MEDUSAE. lxv

funnel cavity of many Medusae, are hollow spaces, more or less conical and lined by the
ectoderm of the subumbrella ; they are always csecal in the aboral bottom of the umbrella
cavity, whilst they open into the coronal cavity of the umbrella by the roundish funnel
openings (" ostia infundibularia "). The adjacent funnels are separated by thin vertical
septa, the mesenteries (" mesenteria " or " mesogonia," ivr). In one group only, the
Pectyllidas (Pis. III.— VIII.). there are eight mesenteries present (four perradial and four
interradial) between eight adradial funnels. Otherwise there are invariably only four
perradial mesenteries between four interradial funnels. The mesenteries or mesogonia
are formed by the four perradial oral corners extending, like wings, in the bottom of the
umbrella cavity and rising in the form of thin folds of the subumbrella. The further
these folds pass towards the outside on the subumbral surface, and the further they pass
downwards on the oral corners, the deeper are the intermediate funnel cavities. In the
Craspedotae, the mesenteries are always thin, delicate membranes, which serve chiefly for
fixing the oesophagus {e.g., among the Anthomedusae in Tiara and Turris, System, taf.
iii., iv. ; among the Trachomedusae in Pectyllis and Pectanthus, Pis. IV., VIII.). In the
Acraspedae, on the other hand, the mesenteries are often hollow, as the central gastral
cavity arches into them like pouches, especially in part of the Lucemaridse ("mesogonial
pouches," " bursae mesenteriales "). The funnel cavities are usually flat and insignificant
in the CubomedusEe (PL XXVI.), but very large and deep in the Peromedusae. In the
Periphyllidae (PI. XXI. figs. 12, 13, ib), they even ascend as far as the point of the
umbrella cone, so that they touch the four interradial taeniola, in the centre point of the
basal stomach (PI. XX. fig. 8, ib). In such a case, the funnels hollow out the entire
length of the four interradial taeniola, so that these solid ridges are transformed into
hollow cones. The four interradial funnel cavities are peculiarly modified in the Dis-
comedusae, where they obtain special importance as "respiratory cavities" or " subgenital
cavities."

§95. Subgenital cavities (" demnia," otherwise also called "respiratory cavities,"
"genital cavities," "umbrella cavities of the reproductive organs," " infundibula sub-
genitalia "). These four peculiar interradial cavities are only found in the order of the
Discomedusaa, where they are in part developed and transformed into peculiarly shaped
hollow spaces. Fundamentally they are merely subumbral funnel cavities, which have
acquired a varied form and function by special adaptation (namely in their relations to
the genitalia). Whilst in the three orders of the Tesseroniae, the four funnels usually
rise as slender, hollow cones, corresponding to the conical or pyramidal form of the high,
vaulted umbrella, in the Ephyroniae or Discomedusse, on the contrary, they extend on
the lower surface of the umbrella, in the form of low pouches, in correlation to its flat
discoid shape. In this order the subumbral wall of the flat, wide gastral cavity is, at
the same time, the place of origin of the reproductive glands, and forms a delicate thin-
walled "gastrogenital membrane" (gg), in which the four interradial (in the Cannostomae

(ZOOL. CHALL. EXP. PART XII. — 1881.) M i

lxvi

THE VOYAGE OF H.M.S. CHALLENGER.

sometimes eight adradial) reproductive bands are developed. The gelatinous support-
ing plate of the subumbrella is often thickened round these genitalia in the form of a
firm, cartilage-like subgenital ring (" annulus subgenrfcalis "). If the genital band
increases considerably and the delicate gastrogenital membrane round it becomes folded
repeatedly, the latter may undergo a double change of position. It either passes below-
through the firm subgenital ring, which does not extend in an equal degree, into the
umbrella cavity, thrusts itself out like a projecting hernia and so forms four pendant ex-
ternal gastrogenital pouches whose cavities are lined by endoderm (" extraversio gonadum,"

Fig. G. Cannorlriza connexa (Discomedusje, Versuridie).

Adradial section, (ug) Gelatinous umbrella, (gc) Central stomach, (gg) Bottom of the central stomach
(gastrogenital membrane with the genitalia, s). (ir) Subgenital porticus. (ah) Braehiferous plate.
(ap) Arm pillars, (cd) Pillar canals, (ga) Buccal stomach, (ab) Oral arms (adradial). (cb)
Brachial canals, (an) Funnel frills (sucking mouths).

System, p. 470), as, for example, in the Cyaneidse (Pis. XXX. XXXL, gg) or, reversed,
the increasing folded gastrogenital membrane, along with the genitalia attached to it,
does not pass through the subgenital ring, but invaginates into the central gastral cavity,
like a replaced hernia (as, for example, in the Aurelidse, System, taf. xxxiii. fig. 7). It
then forms four " inner subgenital cavities," lined with the ectoderm of the subumbrella
(" introversio gonadum," System, p. 470). These are, however, merely flattened funnel
cavities ; the " ostium subgenitale," which may be compared to the " hernial opening "
("porta hernise subgenitalis"), is the narrow opening of the subgenital ring, which leads

REPORT ON THE DEEP-SEA MEDUSAE. lxvii

from the coronal cavity of the umbrella into the four funnel cavities (compare the
detailed description in the System der Medusen, 1879, pp. 467-473). Whilst the four
subgenital cavities remain separate in most Discoinedusae furnished with them (Tetra-
demnise), in part of the Khizostomae (Monodemniae) they are fused in the centre into a
single common " subgenital vestibule."

§ 96. Subgenital vestibule (" portieus subgenitalis, syndemnium," iz). The peculiar
and remarkable hollow space, which we have termed " portieus subgenitabs " (System,
1879, p. 472), is only found in the middle of the subumbrella in two families of the
Rhizostonue, the Versuridse (System, taf. xl.) and the Crambessidse (System, taf.
xxxviii., xxix.), which we have therefore united into the section of the Mono-
demniaB. This central subgenital vestibule has arisen from the four interradial,
subgenital cavities already described, growing centripetally as far as the middle of the
central gastral space, and entering there into immediate communication (PI. XXXII. , and
woodcut, fig. G). The delicate gastrogenital membranes (gz) which form the thin wall
of the invaginated subgenital pouches, touch in the central axis of the gastral cavity and
become fused together ; these points of fusion are then broken through and the four
subgenital cavities, which were originally separated, are consecpiently fused into one.
The cruciform central subgenital vestibule formed in this way (PI. XXXII, fig. 2, iz) is
completely lined by the ectoderm of the subumbrella, and only opens to the outside by
four interracbal portals, the "subgenital ostia" (figs. 1, 7,ig). Its upper wall or the
vestibule roof (" paries portieus gastralis ") is formed by the delicate gastro genital
membrane (gg) which separates it from the overlying gastral cavity (gc) and bears
the four genitalia (s) ; its lower wall, or the " vestibule floor," is formed by the peculiar
brachiferous disk (" discus brachiferus ") from which the eight adradial oral arms of the
Rhizostoinas depend, and in the middle of which below we find the suture of the oral
cross (PI. XXXIL, figs. 2, 6, 7, ah). The two walls are only connected by the four
perradial vestibule pillars (" pilastri," ap). These are four strong gelatinous columns,
placed between the four narrower or wider subgenital openings. The four simple pillar
canals (fig. 2, cd) which represent the only connection between the upper central
stomach (gc) and the lower buccal stomach (go) along with the arm canals (eb)
proceeding from it, run perradially upwards in these columns.

lxviii THE VOYAGE OF H.M.S. CHALLENGER.

IV. GASTEOVASCULAE SYSTEM OF THE MEDUSA.

§ 97. Composition of the gastro vascular system. Of the two large organic systems
composing the body of the Medusa, the gastrovascular system includes the complex of
the vegetative organs, the apparatus for nutrition and reproduction, and is, therefore,
physiologically opposed to the neurodermal system, which forms the complex of the
animal organs. This antithesis is shown histologically in relation to the two primary
germinal layers, as the majority and more important parts of the gastrovascular system
originate from the endoderm (or " vegetative germinal layer "), whilst those of the neuro-
dermal system, on the contrary, originate more usually from the ectoderm (or " animal
germinal layer "). The apparatus of nutrition, formed by the principal intestine (stomach
along with the oral organs) and the radial coronal intestine proceeding from it (vascular
corona or pouch corona), is by far the more considerable and widely differentiated of the
two apparatuses composing the gastrovascular system. The apparatus of reproduction is
much simpler and less differentiated ; it consists solely of the sexual glands or genitalia,
which are developed in the subumbral wall of the gastrovascular system.

§ 98. Hollow space and walls of the gastrovascular system. The entire gastrovascular
system of the Medusae, in spite of its numerous and important modifications in different
groups, shows everywhere one and the same essential type of formation. It appears
everywhere as a more highly developed formation of that simple gastral hollow space,
which is met with in the lowest polyps (Hydra, Clava, &c, among the Hydropolyps ;
Scyphostoma, Spongicola, &c, among the Scyphopolyps). The primitive, perfectly simple
gastral cavity of these oldest polyps is nothing more than the original primitive intestine
(" archigaster, archenteron ") of the gastrsea, which still forms the common ontogenetic
base for the intestinal system in the gastrula of all Metazoa ; its simple opening is the
primitive mouth (" archistoma, blastopores). We distinguish the two walls of this
primitive intestine of the polyps as the aboral calyx wall ("paries calycinalis, calyx")
and the oral peristomal wall (" paries peristomalis, peristomium ") ; the two pass
immediately the one into the other at the margin of the calyx (" margo calycinalis ").
In the Medusa?, the notumbrella corresponds to the calyx on the one hand and the
ccelumbrella to the peristomium on the other ; we, therefore, term the calyx wall of the
gastral space the dorsal wall (" paries umbralis " or " dorsalis ") and the opposite inner
or peristomial wall the ventral wall (" paries subumbralis " or " ventralis "). The
endodermal epithelium of the former is always formed of small flat flagellate cells, that
of the latter of large high flagellate cells (§ 47).

§ 99. Principal intestine and coronal intestine (" axogaster et perogaster "). In all
Medusae the gastrovascular system or intestinal system is divided first of all into two
different principal sections, into a central and a peripheric part. For brevity we shall
term the former the principal intestine, and the latter the coronal intestine. The

REPORT ON THE DEEP-SEA MEDUSAE. lxix

central principal intestine (" gaster principalis, axogaster ") is simple and undivided, its
axis is at the same time the principal axis of the whole body, the umbrella cone (or
centre of the gelatinous umbrella disk) lies at its aboral pole, the oral opening, at the
oral pole. The peripheric coronal intestine ("gaster coronaris, perogaster "), on the
other hand, is always divided by radial septa (or cathamma), into four or more radial
cavities (pouches or canals). The ideal, circular or polygonal boundary line between the
principal intestine, and the coronal intestine is consequently defined by the proximal
ends of the septa or cathamma ; the gastral openings (" ostra gastralia," go) be between
them. These narrow or wider fissures are the only openings by which the central
principal intestine communicates with the divisions of the radially divided coronal
intestine.

§ 100. Cathamma or fused parts. (Fusions of the two walls of the umbrella or of
the dorsal and ventral umbrella ; septa of the gastro vascular system). The only
essential difference between the more simple gastrovascular system of the polyps, and
the more composite system of the Medusae derived from them, consists in this, that the
peripheric part of the latter is divided by radial septa into a number, four at least, of
radial divisions (pouches or canals). These radial pouches and radial canals were
formerly erroneously supposed to be collective evaginations of the central gastral
cavity, which had grown from its margin into the solid peripheric part of the umbrella.
Now we know that these radial hollows have rather arisen from the fusion at definite
points (and first of all, at four interradial points) in the periphery of the simple gastral
space of its two walls (the umbral dorsal wall and the subumbral ventral wall). These
concrescentiae or cathammata (k) form the radial septa of the peripheric gastrovascular
system, between which the remains of the originally simple cavity remains open.
Corresponding to the origin of these septa or cathamma is a double, narrow, fused layer
of endoderm cells, the cathammal plates are found originally in the middle of their solid
gelatinous mass.

§ 101. Cathammal plate ("lamina cathammalis," dk; endoderm lamella, gastral
fused plate, vascular plate). The "endoderm lamella," which on account of its origin and
meaning we term cathammal plate, remains intact in all Medusae in the interior of the
cathamma or septa, and keeps up continous connection between the hollow spaces of the
gastrovascular system, separated by the septa. The cathammal plate consists originally
of a double layer of endoderm cells (PI. XXV. figs. 8, 10) ; the outer or abaxial layer
(the " umbral endoderm lamella," dir) belongs to the dorsal epithelium of the coronal
intestine, and originally lined the concave inner surface of the notumbrella (figs. 8, 9, «</),
whilst the inner or axial layer (the " subumbral endoderm lamella," duf) belongs
genetically to the ventral epithebum of the coronal intestine, and formerly covered the
convex outer surface of the ccelumbrella (figs. 8, 10, zw). Sometimes (as for example,
very distinctly in the firm septal nodes of the Peromedusae, PI. XXV. fig. 8, 10), both

lxx THE VOYAGE OF H.M.S. CHALLENGER.

layers of the cathammal plate remain distinct during the life of the Medusae, and can
even be artificially separated (by suitable pressure) in fine transverse section. This is
however not usually the case. Only a single thin layer of cells is generally visible in
the connective tissue of the cathamma, as the two endoderni plates, which were originally
separate, have become completely fused together. A considerable hardening and
thickening of the two connective plates usually takes place in the cathamma on both
sides of the endoderm plates, and the soft gelatinous tissue is sometimes even trans-
formed into true firm fibrous cartilage (PI. XXV. figs. 8, 10, wg2, zw2).

§ 102. The three principal forms of the cathamma (k). The fusion of the two walls of
the gastral space, which give rise to the cathamma or septa, may appear in three
principal forms, according as they take place in a point, a line, or a surface. In all three
cases the original number of the cathamma amounts to four, and these four primary septa
lie interradially (in the middle between the four primary tentacles), whilst the four
primary radial cavities separated by them lie perradially (in the same meridian planes as
the four primary tentacles). The number of the cathamma may, however, be considerably
increased secondarily (corresponding to variations of the homotypical fundamental numbers
already mentioned, §§ 23-26). In the most simple case when the concrescence takes place
in four points, four septal walls are found (" nodi cathammales," hi) as in part of the
StauromedusEe and in all Peromedusse. The peripheric hollow space of the gastro-
vascular system then appears as a large coronal sinus (" sinus coronaris "), whose division
into four is only indicated by the four small nodes (Pis. XV., XX., XXV.). In the second
case, when the fusion takes place in four lines, four septal ridges (" limites cathammales,"
hw) are found as in most Stauroniedusse, all Cubomedusse, and part of the Disconiedusse
(Cannostomas, and half of the Semostomse, Pelagidse, and Cyaneidse) ; the peripheric
hollow space of the gastrovascular system then forms four radial pouches (" bursas
radiales "), which are separated by the narrow ridges (Pis., XVI., XVIL, XXVI.). In the
third case, when the fusion takes place in four surfaces, four septal plates (" tabulae
cathammales," kt) are formed ; as most Craspedotse, and among the Acraspedae, in part of
the Discomedusa? (in half of the Semostomaa : in the Flosculidse, and Ulmaridae, and in
all the Rhizostomse) (Pis. I., II., XXXIL).

§ 103. Cathammal fissures (" antra septalia "). In the order of the Anthomedusse
(and in this only) there appears frequently, if not universally, a partial dissolution of the
cathamma, and consequently a local separation of the two fused umbrella walls,
by which the peculiar cathammal fissures or septal cavities of these Craspedotse are
originated. In all the cases hitherto observed, they appear as eight adradial cavities,
which are completely closed, and contain a gelatinous fluid. They occupy the greater
part of the subumbrella, are limited below by the umbrella margin, above by the base of
the oesophagus, and are separated from one another by the eight band-shaped longitudinal
muscles of the subumbrella, of which four run perradially on the axial side of the four

REPORT ON THE DEEP-SEA MEDUSAE. lxxi

radial canals, and four interradially between the latter (Pis. I. fig. 2, mi). The abaxial
wall of the cathammal fissures is formed by the thin cell layer of the dorsal endoderm
lamella, which lies on the concave inner surface of the notumbrella, whilst its axial wall
is formed by the supporting plate of the ccelumbrella, of which the ventral endoderm
lamella is lost. The septal cavities are remarkably developed, for example, in Codonium
and Sarsia among the Codonidae, in Tiara and Catablema among the Tiaridae, in Cytacis
and Rathkea among the Margebdae, &c. They have been hitherto erroneously regarded
as ccelome fissures, and placed along with the true body cavity (" cceloma") of the higher
animals. Such a cavity does not, however, exist in the Medusae any more than in other
urticating animals. The septal cavities of these Anthoniedusae rather arise from the two
fused cathammal plates between the eight principal radial Hnes becoming parted second-
arily, and only remaining fused in these eight lines. From this we see most clearly that
the cathammal plates pass continuously into the endoderm layer of the radial canals, where
these touch the lateral margins of the septal cavities. On the other hand, the cathammal
plates are completely separated from the ectoderm of the umbrella, externally by the
gelatinous body of the notumbrella, internally by the supporting plate and muscular
plate of the ccelumbrella.

§ 104. Gastrovascular system of the Craspedotae and Acraspedae. The two sections
of the class Medusae show differentations, which are perfectly analogous, in the forma-
tion of their gastrovascular system, but still present, in spite of all similarity, a constant
and therefore very essential difference. In all the Acraspedae, movable gastral filaments
(" filamenta gastralia"), or "internal gastral tentacles," are found on definite places,
whilst these are never present in the Craspedotae. In the Acraspedae there are at least
four of these gastral filaments, which are regularly distributed interradially. They are,
however, generally very numerous (usually over a hundred, often over a thousand), and
arranged in such a manner as to form four interradial groups of filaments (" phaeelli ").
The movable filaments of these phaeelli are sometimes arrayed in a single row, some-
times in several rows ; they are sometimes simple, sometimes branched, and always
consist of a solid, cylindrical or band-shaped gelatinous filament (a process of the fulcrum)
covered with endodermal epithelium. The phaeelli of the Acraspedaa (or "Phacellotas")
are, therefore, of great phylogenetic importance, as indications of them already exist in
the polyp nurses of this section. The Scyphopolyps, from which the Acraspedae are
descended, have all four interradial gastral ridges or gastral taeniola on the inner
surface of the wall of the stomach, and from these the " filaments " of the " Scyphornedusae "
are developed.

§ 105. Taeniola or gastral ridges (" taeniola gastralia, " ft). The phylogenetic hypo-
thesis (§ 6) that the two sections of the class Medusae have arisen independently of
one another from two different groups of polyps is fundamentally supported by the
fact that the important difference in their gastral formation mentioned above (their being

xxii THE VOYAGE OF H.M.S. CHALLENGER.

with or without gastra] filaments) is already present in the corresponding sections of
the class Polypi, from which the Medusae are descended. The lower Hydropolyps (from
which the Craspedotae or Hydromedusae) are descended, never possess the four
characteristic interradial taeniola of the Scyphopolyps, although similar longitudinal
ridges of the inner gastral wall (more irregular in number, form and position) are also
found in some Craspedotae (Tubulariae, &c). On the other hand, the higher Scyphopolyps
from which the Acraspedae (or Scyphopolyps) are descended, are all originally dis-
tinguished by the presence of the four interradial taeniola (also called " gastral walls,
mesenteric swellings, longitudinal ridges, longitudinal walls, longitudinal swellings of
the inner gastral wall "). These longitudinal gastral ridges (which are also universal
among the corals as the so-called mesenteric filaments or mesenteric bands, more
properly " gastral bands "), appear originally in the Scyphopolyps as four interradial rib-
like thickenings of the gelatinous supporting plate or fulcral lamella ; they project from the
inner surface of the gastral wall freely into the gastral cavity, and in this way divide its
peripheric hollow space into four perradial niches or grooves (PI. XV. figs. 2, 3, 7, 8, ft ;
PI. XVII. figs. 13, 14, ft). They usually contain a longitudinal muscle (PI. XVII. fig. 13,
on) and are always covered by endodermal epithelium.

§ 106. Dorsal and ventral taeniola (" taeniola notumbralia" and " ccelumbralia "). As
the taeniola or gastral ridges of the Scyphopolyps are developed over the whole extent of
their gastral surface and originally (in Scyphostoma, Sjiongicola, &c.,) run from theaboral
to the oral pole, from the foot plate of the peduncle to the margin of the mouth, we
must divide them into two principal sections, a notumbral part and a ccelumbral part ;
these two pass into one another at the peristom margin or calyx margin of the
Scyphopolyps (corresponding to the umbrella margin of the Scyphomedusae). The
notumbral or dorsal tamiolum reaches from the aboral base or point of the cone to the
peristomial margin or umbrella margin, and is a ridge-like thickening of the calyx or
" notumbrella " (covered with dorsal endoderm). The ccelumbral or ventral taeniolum
reaches from the peristomial margin to the umbrella margin, and is a ridge-like thickening
of the " ccelumbrella " covered with ventral endoderm. The cathamma or septa of the
Acraspedae (k) are formed at the points where the two portions of the taeniola, the dorsal
and the -ventral ridges, come in contact and become fused.

§ 107. Gastral filaments (mesenteric filaments, internal gastral tentacles, f). The
characteristic " gastral filaments " of the Acraspedae, which are universally present in
them and totally wanting in the Craspedotae, are originally papillae or excrescences of
the taeniola. We immediately perceive from such Acraspedae, in which the taeniola
persist in their original form (especially Stauromedusae) that the filaments are originally
parts of the taeniola (Tesserantha, PI. XV. ; Lucemaria, PI. XVI., XVII). This is also the
case in the strobilation of the Discornedusae, where the separate parts of the four strobila
taeniola are immediately transformed into the four primary filaments of the Ephyrula

REPORT ON THE DEEP-SEA MEDUSAE. lxxiii

Medusa which detaches itself from the strobila (fig. K,fp.). Corresponding to this origin,
each filament consists of a solid (sometimes cylindrical, sometimes flattened) gelatinous
filament or fulcral papilla, which is simply a papillose or digital excrescence of the
supporting plate of the taeniolum. The endodermal epithelium of the latter passes directly
on to the filament and consists partly of flagellate cells, partly of glandular cells (calyx
cells) to which thread cells and (perhaps universally 1) delicate muscular epithelial cells are
often added (comp. PI. XXII. figs. 23-26). The movable gastral filaments of the Medusae
are usually simple, more rarely branched dichotomously or even pinnated (PI. XXVI.

108. Phacelli, or groups of filaments. It is only in the simplest and oldest genera of
the Acraspedas {Tessera, System, taf. xxi.; Epliyra, taf. xxvii.) that the filamental

Fig. H. Procharagma proiotypus (Cubomeduss, Charybdeidse).

Horizontal transverse section below the stomach, whose subumbral wall (gc) is completely visible, in
the middle, the oral opening (a) with the four perradial oral lobes (al). The gastral filaments (/)
are placed upon the four interradial pyloric valves (gy). (gw) Subumbral wall of the two gastral
pouches (bp). (s) Genitalia, (ug) Gelatinous substance of the umbrella.

products of the taeniola are limited to the simple gastral filaments (lying immediately on
the four interradial cathammal nodes), and also in the young larvae of other Acraspedae,
e.g., in the Ephyra larvae of Aurelia, there are at first only four such simple filaments
(fig. A, fix). In all other Acraspedae numerous filaments are present, which form separate
" groups of filaments " or phacelli. There are always originally only four interradial
phacelli (fig. D, f) ; these have arisen by division of the four originally simple fila-
ments, or by repeated production of filaments from the taeniola. Instead of the
four interradial, there are often eight adradial phacelli, which have arisen by bifurca-
tion of the former, and are, therefore, more accurately, four pairs of phacelli. These are
usually united in pairs at the proximal end, whilst they diverge at the distal end. The
distal division of the eight phacelli is strongest in the Periphyllidaa, where the four

(ZOOL. CHALL. EXP. — PAET XII. — 1881.) M k

lxxiv

THE VOYAGE OF H.M.S. CHALLENGER.

taeniola are hollowed to their aboral ends by the four endodermal conical funnel cavities
of the subumbrella (Pis. XXI., XXII.). The phacelli sometimes form straight lines,
sometimes arches more or less waved and have often a complicated course as in
Periphyllida3 (Pis. XXL, XXII). The tseniola may form phacelli in all the three
chambers of the principal intestine ; in most Acraspeda? the formation of them is limited
to the central stomach. In the Peromedusse and in some Stauroniedusse, they are

Fig. I. Pcrkolpa quadrigata ( Peromedusce, Perieolpida?).
Interradial section. (ug) Gelatinous umbrella, (cs) Coronal sinus (cs' proximal part, cs- distal part).
(s) Genitalia, (hi) Cathammal nodes, (go) Gastral openings, (md!) Deltoid muscles, (roc)
Coronal muscle, (ft) Tamiola. (gb) Niches of the basal stomach between the tamiola. (gy)
Pylorus, (gc) Central stomach, (gp) Palatine opening, (ga) Buccal stomach (oesophagus), (am)
Oral margin, (oi) Sense clubs (interradial). (oa) Ampulla; at their base, (bu) Horseshoe-shaped
canal of the marginal lobes (la). (M) Peronia between the two limbs of the canal (bl). (tp)
Tentacles (perradial).

strongly developed in the basal stomach, in the Cubomedusse and Peromedusae in the
buccal stomach : their terminal processes are the barbous filaments of the Peromedusae
(PI. XX. figs. 9-11, af). In the Discomedusse, the phacelli assume a definite topo-
graphical (and perhaps also physiological) relation to the reproductive glands ; they he
there on the inner surface of the subumbral gastral wall, on the inner or axial margin
of the frill-shaped, folded genitalia, whose winding course they follow.

REPORT ON THE DEEP-SEA MEDUSAE. lxxv

§ 109. The three chambers of the principal intestine ("gaster principalis"). The
central principal intestine of the Medusae is rarely perfectly simple, it is usually
divided more or less distinctly into two or three sections or chambers, lying one above
the other in the principal axis of the body. The lowest of these is the oesophagus or
buccal stomach (" gaster buccalis," fig. I, go), which contains the oral opening at the oral
pole of the principal axis. The middle chamber is the principal cavity or central
stomach (" gaster centralis," gc). The third or uppermost section is the peduncle tube or
basal stomach (" gaster basalis," gb), which ends caacally at the aboral pole of the principal
axis. The central stomach communicates with the basal stomach, below by the palatine
opening (" porta palatina," gp), above by the pyloric opening (" porta pylorica," gy) ;
besides these there are usually gastral openings (" ostia gastralis," go), in the lateral
walls of the central stomach, by which the latter communicates with the radial chambers
of the coronal intestine. All the three stomachs are well developed in many Medusas of
both sections (namely Anthomedusae and Peromedusae) ; the uppermost (basal) stomach
has, however, usually undergone retrograde formation. In the majority of Medusas, the
buccal stomach is the longest, the central stomach the broadest of the three chambers,
whilst the basal stomach is the smallest or has disappeared. All the three chambers can
be already distinguished in many polyps (both Hydropolyps and Scyphopolyps). The
buccal stomach of the Medusae corresponds to the freely projecting proboscis (" tubus
oralis ") of the polyps, the central stomach to their peculiar " calyx stomach " (" cavitas
calycina "), and the basal stomach to their peduncle tube (" tubus peduncularis ").

§ 110. Buccal stomach or oesophagus ("gaster buccahs," also termed "tubus oralis,
proboscis, manubrium," go). The buccal stomach or oesophagus (go) is the first and
lowest of the three chambers of the principal intestine. It is evolved from the
"oesophagus" or "proboscis" of the polyps, and shows extraordinary diversity of size,
form, and differentiation. It originally bears the oral opening (aa) at the oral pole of its
vertical axis, whilst it opens at the aboral pole of the axis into the central stomach (gc)
by the palatine opening (gp) ; the latter is sometimes sharply defined, sometimes
indistinct. The palatine opening is usually found in the middle of the subumbrella, so
that the oesophagus hangs freely from the latter into the umbrella cavity. In the
majority of the Medusae, the oesophagus is the longest, but not the broadest of the three
gastral chambers. It is developed exceptionally strongly in the Anthomedusae, Tracho-
medusae, Peromedusae, and Discomedusae, whilst in the other orders it is usually weak
or often quite rudimentary. The oesophagus rarely forms a simple cylinder without
radial division (as in the Codonidas, System, taf. i., ii. ; and in many Narcomedusae,
System, taf. xix. xx.). In the majority of Craspedotae, and in all Acraspedae, the
oesophagus is prominently quadrilocular, as the four perradial buccal ribs (" costae
orales," ab) have a tendency to centrifugal growth, project towards the outside, and
become prolonged into the midribs of the oral lobes, whilst the four intermediate buccal

lxxvi THE VOYAGE OF H.M.S. CHALLENGER.

columns (" columnse orales," ae) have a tendency to centripetal growth, project inwardly,
and end below in the archings of the oral margin.

§ 111. Oral opening (" actinostoma, apertura oris, osculum," eta). In all Medusae, the
mouth is originally a simple, usually quadrate or cruciform opening at the lower end of
the buccal stomach. Its margin, however, rarely continues perfectly simple, different
organs being commonly developed from it, of which the four perradial oral lobes and the
oral arms, which have arisen from their prolongation, are by fax the most usual and the
most important (§ 113). The terminal oral opening itself in the Medusae usually shows
the same characteristic cross figure as the transverse section of the oesophagus, the
typical oral cross ("stomostaurus"), with four limbs projecting perradially and four
intermediate angles projecting interradially (PL I. figs. 2, 4 ; PI. XV. figs. 5, 6, &c). The
perfectly constant position of this oral cross is very important for the orientation of the
transverse axes. The free oral margin or the margins of the oral arms are usually
strongly armed with thread cells, which are often placed in special regularly distributed
groups (PL I. fig. 4). As the gelatinous supporting plate below these groups of thread
cells is arched conically or hemispheroidally, oral urticating papillae or urticating
knobs are formed (oral papillae, "papillae orales," e.g., in Pelagia). If these
grow in length, they become developed into the tentacle-like, cylindrical movable
filaments, which serve, like the true (marginal) tentacles, both for feeling and for seizing
upon prey. The structure of these oral tentacles or oral fingers (" digitella ") completely
resembles that of the inner "gastral tentacles," or gastral tentacles (§ 107) with which they
were formerly often confounded (" oral filaments "). But the epithelium, which covers
the solid gelatinous axis of the two analogous organs, belongs to the endoderm in the
gastral filaments and to the ectoderm in the oral digitella. In many Semostomas {e.g.
Aurelia) and in all Rhizostomae, a large number of digitella beset the margins of the oral
arms. The oral styles (" stomostyli ") are apparently similar, but essentially different
organisms. They are developed principally in the Anthomedusae, where they characterise
the families of the Margelidae and Dendronemidae (System, p. 70, taf. v.-vii.). In
structure they completely resemble the solid marginal tentacles, and consist of a
cylindrical axis, formed of a single row of endodermal chordal cells (PL I. fig. 5, d) ;
these are separated by a firm, elastic supporting plate (z), from a thin muscular plate (m),
whose longitudinal fibres are connected with the ectodermal epithelium (q) ; the free distal
end bears a spheroidal urticating knob (m). There are originally only four simple oral
styles present at the four perradial oral angles (Oytceis, Lizusa, System, taf. v.-vii.).
The oesophagus is sometimes prolonged secondarily into a long pendant proboscis, so that
the oral styles, which were originally terminal, are found at its base (Thamnostylus,
PL I. fig. 1 ; Limnorea, Nemoposis, System, pp. 86, 92, taf. v.).

§112. Buccal pouches and oral columns ("bursae buccales" et " columnae orales").
In some Medusae of both sections, the thin, extensible walls of the oesophagus are

REPORT ON THE DEEP-SEA MEDUSAE. lxxvii

extended into large evaginations, which project centifrugally to the outside (into the
umbrella cavity), and may be compared from their form and function to the buccal pouches
of many mammalia (bb). We term the thicker parts of the oral wall, which project
inwardly centripetally between them, the oral columns {etc). The buccal pouches are
most strongly developed among the Acraspedae in the Peromedusae, where they form four
powerful perradial archings outward of the large buccal stomach and appear inflated
hemispheroidally or even spheroidally (as in Periphylla, Pis. XVIII.-XXV., bb). Each
single inflated buccal pouch is sometimes more voluminous here than the whole central
stomach. The four thick intermediate, interradial oral columns (ac) project internally
with their axial surfaces extended like wings, in such a way, that the adradial side
spaces of the four buccal pouches form special niches or wing pouches behind these oral
wings (ad, comp. also System, p. 405, taf. xxiv. fig. 14). Among the Craspedotse the
buccal pouches are most strongly developed in the Trachomedusse, whose oesophagus is
often highly extensible. Pedis (PI. V. figs. 2-5 ; PL VI. fig. 11) has sixteen subradial
buccal pouches, which are developed in pairs from four perradial and four interradial
evaginations of the oesophagus, and are separated by eight adradial subumbral oral funnels
(PL V. fig. 5, id). Other Trachomedusse have eight adradial buccal pouches (formed by
the division of four perradial).

§ 113. Oral lobes and oral arms (" lobi orales," al; "brachia oralia," ab). In the
majority of Medusae the oral margin is not simple, but its four perradial corners are
prolonged into four leaf-shaped oral lobes or oral arms, between which four interradial
oral sinus or oral incisions (" sinus orales ") project internally. The four oral lobes
have usually the shape of a thin oral leaf, whose delicate margins are folded more or less,
often very daintily, whilst a strong midrib projects in the middle (as distal end of the
perradial oral corner). The supporting plate of this midrib is often thickened gelatinously,
shaped like a groove, concave outside and convex inside. The axial inner side of the oral
lobes is always covered by endoderm, its abaxial outer side by ectoderm (comp. in the
System, 1879, the oral lobes of the Anthomedusse, taf. iii., iv. ; of the Leptoniedusas,
taf. viii., x., xi., xiii. ; of the Trachomedusse, taf. xvi., xviii. ; of the Narcomedusse, taf.
xix., xx. ; of the Stauromedusaa, taf. xxi., xxii. ; of the Peromedusae, taf. xxiii., xxiv. ;
of the Cubomedusse, taf. xix., xx. ; of the Discomedusse, taf. xxvii., xxx., xxxii., xxxiii.).
The oral lobes are termed " oral arms," if the four perradial oral lobes are very large, and
the intermediate oral sinus so deep that they divide the oesophagus almost, or completely
into four parts. These oral arms attain an extraordinary development in the order of the
Discomedusse, in the sub-order of the Semostomse, namely, in the " pennon-mouthed "
Pelagidse (System, taf. xxxi.), and Cyaneidse (System, taf. xxx.). In most Cyaneidae
they resemble delicate curtains with numerous folds, which are frequently larger than the
whole umbrella, and are therefore termed " oral curtains" (Pis. XXX., XXXL).

§ 114. Quadripartite mouth of the Rhizostomse. The third and last sub-order of the

lxxviii

THE VOYAGE OF H.M.S. CHALLENGER.

Discomedusse, the rich group of the Ehizostomse (System, 1879, pp. 464, 560, taf.
xxxiv.-xl.) is distinguished by an extremely peculiar development of the mouth, unique
of its kind in the whole animal kingdom. This has arisen phylogenetically from the
second sub-order, the Semostomse (loc. cit., taf. xxx.-xxxiii.), as the latter is derived from
the Cannostomse, loc. cit., taf. xxvii.-xxix.). This plainly points out the ontogeny of the
Ehizostomse, which are Cannostomse in their earliest stage, then become Semostomae,
and are finally transformed into Ehizostomse. The simple quadrangular oesophagus of
the Cannostoinse first shows four delicate, frilled oral lobes at the oral margin (Ephyridse,

<&: ir

x

P^^^^^ 1

J

w^^ .J^^^ B^^ HUB HHw& ^^H

(

t^Z*

1^h

w

i §

| rS

^mMM

m 'm

Wmk:

m

^■WLr

mm m

at

BtT JvJa wMK

5©

■^w^ «mw-~a* ]

Fig. K. Cannorhiza connexa (Discomedusse, Versuridre).

Adradial section, (ug) Gelatinous umbreUa. (gc) Central stomach, (gg) Bottom of the central stomach
(gastrogenital membrane with the genitalia, s). (ir) Subgenital porticus. (ah) Brachiferous plate.
(ap) Arm Pillars, (cd) Pillar canals, (ga) Buccal stomach, (ab) Oral arms (adradial). (cb)
Branchial canals, (am) Funnel frills (sucking mouths).

System, taf. xxvii. ; Flosculidse, taf. xxvii.). As these increase considerably in size, and
the intermediate four interradial oral sinuses take the form of deep incisions, they are
transformed into four strong, perradial oral arms, which are developed in most Ulmaridse
into long oral pennons (System, taf. xxxi.-xxxiii). In one Ulmarid (Aurosa, System,
taf. xxxiii. figs. 7, 8), the four perradial oral arms are bifurcated at the distal end into
two lobes. In this way, the eight adradial oral arms are formed, which are present in
all Ehizostomse, and which are only connected in pairs at their bases ; where they are
usually connected with four strong perradial oral pillars (ap)- Whilst the thin leafdike

REPORT ON THE DEEP-SEA MEDUSAE. lxxix

margin of the arms are strongly frilled, their thick midrib is deepened into a groove on
the concave endodermal side (oral grooves, " sulci orales," az). The separate folds of the
strongly frilled oral margins are laid against one another in such a way that the opposite
and contiguous endodermal surfaces of the groove-shaped folds become fused at the points
of junction, and transformed into short canals. They open freely on the outside by a
funnel at the distal end, towards the inside into the perradial oral groove by a fissure.
By fusion of its margins, this oral groove is transformed into a canal (brachial canal).
Finally, as the central oral opening becomes fused between the bases of the four oral
pillars, and closed by a cruciform " oral suture " (PL XXXII. fig. K), they are physio-
logically represented by the numerous " frill funnels " (funnel mouths or sucking mouths).
This polystome of the Bhizostomae, therefore, finally rests upon the division into many
parts of the originally simple mouth ; the central part becomes fused whilst it is replaced
by numerous peripheric oral funnels. The fusion of the oral frills is therefore dependent
upon endodermal concrescence or the formation of cathamma, as in the septa between the
subumbral radial divisions of the coronal intestine. The endodermal cathammal plate
(§ 101) is persistent in both cases. The funnel frills of the BhizostomEe, whose freer oral
margin is usually thickly beset with countless digitellae, may produce varied appendages.
Thus, urticating clubs are formed by annular fusion (" concrescentia annularis"), (e.g.,
Cassiopea, Cotylorhiza, System, taf. xxxvii.), whilst the urticating scourges are formed
by fissure-shaped fusion ("concrescentia longitudinales "), ("e.g., Cephea, Lychnorhiza,
System, taf. xxxiv.).

§ 115. Palate or palatine opening (" palatum, porta palatina," gp). This is the name
applied to the circular constriction by means of which the oesophagus in most Medusae is
more or less clearly separated from the central stomach. In many cases it is insignificant
or almost obliterated, whilst in others it appears distinctly as a slender neck (comp. in the
System, for example, the Anthomedusae, taf. iii., iv., vii. ; the Leptomedusae, taf. x., xii.,
xiii. ; the Trachomedusae, taf. xvi., xviii. ; the Narcomedusae, taf. xix., xx. ; the Stauro-
medusae, taf. xxi., xxii. ; the Peromedusae, taf. xxiii., xxiv.; the Cubomedusae, taf. xxv.,
xxvi. ; the Discomedusae, taf. xxvii., xxviii., xxx., xxxi., &c). The palatine opening is
usually the narrowest part of the principal intestine, and in many cases appears capable
of closing voluntarily, so as to debar the water from passing in from the buccal to the central
stomach. The form of the palatine opening is usually distinctly cruciform ; the four per-
radial limbs of this "palatine cross" forming the upper terminal portion of the " oral cross,"
project centrifugally towards the outside, their concave abaxial ends being often depressed
into a groove, (palatine groves " sulci palatini,") gs, PI. XX. fig. 11). They are sometimes
supported by a strong, node-like thickening of the fulcral lamella (palatine nodes, " nodi
palatini," gk, PI. XX. figs. 9, 10) The four interradial palatine lips ("labia palatina" #Z),
which project centripetally inwards, lie between the palatine nodes (System, taf. x. fio-.
6 ; taf. xxiv. fig. 14 ; taf. xxviii. fig. 5, taf. xxxi. fig. 3) ; the gelatinous fulcral plate

lxxx THE VOYAGE OF H.M.S. CHALLENGER.

of the latter is often as strongly thickened as that of the former, and frequently becomes
as hard as cartilage, especially in the larger Acraspedee. In many Cyaneidse it forms
a thick palatine ring as hard as cartilage (" annulus palatums," PI. XXX. figs. 30, 31, aw).
The Cubomedusas Chirodropus is distinguished by the development of its palatine lips,
into four intcrradial palatine valves (" valvulse palatinse," System, p. 429, taf. xxvi.
figs. 3, 4, k) ; they resemble the semilunar valves of the human heart in form, have their
concave side turned to the central stomach, which they can close completely and so
prevent any communication with the oesophagus.

§116. Central stomach or central cavity (" gaster centralis, cavitas centralis, gc).
The second and middle chamber of the three chambers of the axial principal stomach, the
central stomach, is homologous in the free-swimming Medusa, with the central calyx
cavity, or the true stomach of the sessile polyps. It is separated from the buccal
stomach above by the palatine opening (" palatum "), from the basal stomach, below by
the pyloric opening ("pylorus"). In the side walls of the central stomach there are
four perradial (rarely more) openings ("ostia gastralia") by which it communicates with
the surrounding pouches or canals of the coronal intestine. The general form and
relative size of the central stomach varies most remarkably, and is often difficult to define.
It usually forms the widest and broadest of the three chambers of the principal intestine,
whilst the buccal stomach is the longest. The central stomach is sometimes entirely en-
closed in the gelatinous substance of the umbrella, sometimes not ; in the former instance
its horizontal axis is usually considerably larger than its vertical, in the latter instance it
is usually the reverse. We can therefore generally distinguish two principal forms of
the central stomach, which are, however, connected by numerous intermediate forms and
cannot be sharply defined, — the high obelisk stomach, and the flat lens stomach. The
high obelisk stomach (" gaster centralis obeliscus ") has usually the form of an obelisk or
a truncated quadrate pyramid (Pis. XV.-XXIV.) ; the palatum forms its lower base, the
pylorus its upper base, the four perradial gastral ostia correspond to the four angles of
the obehsk, the four interradial side walls of the central stomach or the obelisk plates
(" tabula? obelisci," gz) to the four sides. The vertical axis of the obelisk stomach is
usually larger than the horizontal diameter. It usually hangs freely in the umbrella
cavity and is often fastened to a shorter or longer " gastral peduncle " (" pedunculus
gastralis," «s, see above) ; this is the case in the majority of the Anthomedusas (System,
taf. iii., iv.; vii.), of the Leptomedusae (System, taf. xi.-xv.), and of the Trachomedusse
(System, taf. xvi.-xviii.), also among the Acraspeda? in most Tesseronise, in the
Stauromedusse as well as the Peromedusse and Cubomedusae (System, taf. xxi.-xxvi.). In
many cases four perradial mesenteric folds or mesogonia serve to fasten the freely hanging
obelisk stomach to the subumbrella ; the Pectyllidae are distinguished by eight such
mesogonia (four perradial and four interradial) (Pis. III.-VTIL). The second principal
form of the central stomach, the flat lens stomach (" gaster centrabs lenticula ") only

REPORT ON THE DEEP-SEA MEDUSAE.

lxxxi

predominates in two orders, in the Narcoruedusse among the Craspedotae (System, taf.
xix., xx.), and in the Discomedusse among the Acraspedse (System, taf. xxvii.-xl.). The
horizontal axis of the central stomach is usually much larger than the vertical, and the
lenticular flat central stomach does not project at all, or only slightly into the subumbral
umbrella cavity ; its peripheric margin is cut away like a lens. As the basal stomach,
and, consequently, the pylorus, are wanting, the flat upper wall or cover of the stomach is

Fig. L. Cannorhiza conncxa (Discomedusa;, Versuridie).

Subliminal view of the umbrella. The arm disk with the eight oral arms is removed, as the four per-
radial arm pillars (ab), which connect the umbrella disk and the arm disk, are cut through, (oi)
Interradial sense clubs. (um) Umbrella margin (turned over inwards), (s) Genitalia, wx)
Gelatinous cross of the gastrogenital membrane, (gy), (gh) Peripheric limbs of the gelatinous
cross, (ug) Peripheric umbrella corona, (cc) Coronal canal, (cd) Pillar canals, (ca) Adradial
canals, (ft) Interradial canals, (cp) Perradial canals.

formed immediately by the discoid gelatinous substance of the umbrella ; the lower wall
or bottom of the stomach is supported by a thick gelatinous plate of the subumbrella, and
communicates in the middle through the palatum with the oesophagus (Pis. XXVI.-
XXXI. ).

§ 117. Cruciate chambers and cruciate columns ("camera? cruciatae" et "columnar
cruciatas "). In many Medusae of different groups the quadrate central stomach assumes

(ZOOL. CHALL. EXP. PART XII. — 1881.) M I

lxxxii THE VOYAGE OF H.M.S. CHALLENGER.

a more or less distinct cruciate form, as four perradial archings outwards alternate with
four interradial archings inwards in the same fashion as in the oral cross and the palatine
cross (PI. XXXII. figs. 3, 4). We therefore term the four perradial limbs of the central
gastral cross, the cruciate chambers, and the four interradial gelatinous pieces of the
umbrella, projecting centripetally between them, the cruciate columns. The latter
correspond morphologically to the interradial centre of the obelisk plates, and at the
same time to the central section of the taeniola. The cruciate form of the central stomach
is prominently developed in many Anthomedusae (System, taf. iv., figs. 7, 9), the
Leptomedusa (System, taf. viii. fig. 6), Stauromedusae (System, taf. xxi., xxii.), and
Discomedusae (System, taf. xxviii., xxix., &c). The cruciate chambers are most strongly
developed in the sub-order of the Rhizostomae, in which they are often considerably
larger than the central cruciate cavity (System, taf. xxxvii.-xl.). The cruciate chambers
attain special importance, when the reproductive glands are mostly or entirely developed
in them ; in such cases they are sometimes developed into independent genital pouches
as in many Anthomedusae and Discomedusae.

§118. Gastral fissures (" ostia gastralia," go). The communication between the
central principal intestine and the peripheric coronal intestine in all Medusae takes
place exclusively by means of the radial " gastral fissures," which are usually limited to
the side-wall of the central stomach. The number of these gastral ostia corresponds to
that of the chambers of the coronal intestine which open into the central stomach, and
therefore amounts to four in the majority of the Medusa?. Four secondary interradial
gastral ostia are, however, often added to the four primary perradial, and sometimes also
a larger and usually variable number of succursal gastral fissures. The latter vary
extremely in form and size; In general, the gastral ostia naturally form large and wide
fissures in those Medusae, whose coronal intestine consists of broad pouches and narrow
septa, therefore in most Acraspedae, and, for example, in the Tesseroniae, the Cannostomae,
and in the Typhlocannae among the Semostomee (Pis. XV.-XXXL). On the other hand
the gastral ostia form small and narrow holes in those Medusas, whose coronal intestine
is composed of narrow canals with broad septa ; therefore in most Craspedotae, both in
the Cyclocannae and Rhizostomae, among the Discomedusae (Pis. I. -VIII. , XXXII. ).
According to their position, the gastral openings sometimes lie vertically, sometimes
obliquely, sometimes horizontally. Most Tesseroniae are distinguished by very large,
wide gastral ostia ; they lie vertically or subvertically in most Peromedusae, and in one
part of the Stauromedusae (Pis. XVI. -XXV.), whilst they are horizontal or sub-horizontal
in most Cubomedusae (PI. XXVL), and in another part of the Stauromedusae ; in all
cases the margins of the gastral ostia are entirely or for the most part edged with
phacelli (PI. XVII. fig. 21). Sometimes, namely in the Cubomedusae, there are valves
by which the gastral ostia can be closed. The small Narcomedusae group of the
Solmonetidae, in which the entire coronal intestine is obliterated, is distinguished by the

REPORT ON THE DEEP-SEA MEDUSAE. lxxxiii

entire want of the gastral ostia (Solmaris and Solmoneta, System, taf. xix. figs. 10-12 ;
taf. xx. figs. 7-10). In consequence of this retrograde formation in Solmaris, the whole
gastrovascular system is reduced to a simple, lenticular, central stomach.

§ 119. Pylorus or gastral opening ("porta pylorica, " gy). In most Medusae in
which the basal stomach is developed, the latter appears more or less sharply defined
from the central stomach by a circular constriction. We term this the gastral opening or
pylorus ; it comports itself above in the same way as the palatine opening does above, but
is wanting, of course, in all Medusae in which the basal stomach has undergone
retrograde formation. In the Anthomedusse, the only Craspedotae which have a
basal stomach or apical canal, the pyloric opening is a simple, circular stricture, and only
of special interest in the Cladonemidae, as it perhaps corresponds to the " funnel opening "
which separates the ectodermal " oesophageal stomach " from the endodermal ;t funnel
cavity" in the Ctenophorae (Ctenaria, System, p. 107, taf. vii.). The pylorus, like the
basal stomach, is usually wanting among the Acraspedae, in the Discomedusae or
Ephyroniae, whilst it is usually very pronounced in the Tesseroniae. In the Stauromedusae
the plyoric stricture is sometimes obliterated, sometimes deeply. inserted (taf. xv. figs. 2,
3, AB). In the Peroniedusae it is sharply defined by the proximal margin of the
coronal sinus (PI. XXI. figs. 12, 13, CD). Special " pyloric valves " (" valvulae pyloricae ")
which project inwardly from the four interradial angles of the quadrate stricture and
from the bottom of the four small pyloric pouches (" bursae pyloricae, " by), are often
developed in the Cubomedusae and in the Cannostomae.

§ 120. Basal stomach or basal tube (" gaster basalis," vel " tubus cupolaris," also
termed " peduncle canal, peduncle tube, apical canal, or cone canal," gb). The basal
stomach is the third and uppermost of the three principal chambers of the central
principal intestine ; it is separated by the pylorus from the central stomach, and
corresponds to the peduncle tube or peduncle canal of the stalked polyps. It is lost in
the majority of Medusae, and only preserved by inheritance in a few groups. We only
find it among the Craspedotae in part of the Anthomedusae, where it appears as a simple,
narrow canal, which traverses the apical process of the umbrella cone, ending caecally in
the point of the latter ; the stomach of the germinating Anthomedusae was originally
connected by this peduncle canal with the stomach of the nursing Tubularia polyp (comp.
System, p. 5, taf. i., ii., iv., vii.) ; in many Cladonemidae it is inlarged into a spacious
apical cavity (" cavitas cupolaris "), which is sometimes occupied by the young brood
(Pteronema, Eleutheria). The enlarged pyriform apical cavity of the Cladonemidae is of
great interest as it probably corresponds to the funnel cavity of the Ctenophorae (System,
pp. 99, 107, taf. vii., Ctenaria), Among the Acraspedae the basal stomach is usually
entirely wanting in the depressed Discomedusae or Ephyroniae, whilst it is constant
and usually highly developed in the three other orders, the highly arched Tesseroniae,
It appears in the Stauromedusae (in the Tesseridae, PI. XV., and Lucernaridae, PI. XVI.,

Lxxxiv THE VOYAGE OF H.M.S. CHALLENGER.

XVII.) as a shorter or longer quadrilateral canal which traverses the apical process or
peduncle of the umbrella, and terminates csecally at its aboral end. The four interradial
tamiola (ft) divide the peduncle canal into four perradial semicanals (peduncular grooves
or niches of the basal stomach, "semicanales basales," gn., PL XVI. fig. 13). In some
Lucernaridse these become four separate peduncular canals (" canales basales "), as the
four interradial tamiola met in the axis of the stomach where they are fused into a central
column (" columella "). In the Peromedusse, the basal stomach is always highly
developed, is sharply defined from the central stomach by the pyloric ring (which
corresponds to the upper margin of the " coronal sinus "), and is divided by the projecting
tseniola into four niches or semicanals. The division into four of the conical basal
stomach is more strikingly defined if the four tamiola are transformed into hollow cones
by the subumbral funnel cavities, which in Periphylla penetrate to the point of the cone
(PI. XXI. figs. 12-18, ib.). In the Cubomedusse the basal stomach is sometimes fused
with the central stomach, sometimes separated from it by a pyloric stricture, in which
case it forms a very flat, low, quadrate cavity ; its four interradial angles are sometimes
extended into four low, triangular pyloric pouches (" bursse pyloricse, by), which are
separated from the central stomach by four projecting pyloric valves (System, p. 430).

§ 121. The two principal forms of the coronal intestine (" gaster coronaris,"
" perogaster "). Under the term "coronal intestine," we include the whole peripheric
gastrovascular system of the Medusas, which surrounds the central principal intestine
like a garland, and only communicates with it by the gastral openings. Although it
has essentially the same formation in both sections of the class, we may assume that it
has been developed independently in both sections. The fusion of the two walls of the
peripheric gastral space, by which the radial chambers of the coronal intestine originate,
shows an essential difference in the two sections. In the Craspedotse, which are descended
from Hydropolyps (without tseniola), the concave inner surface of the not-umbrella
or dorsal umbrella, is fused almost throughout its entire extent with the convex outer
surface of the ccelumbrella or ventral umbrella, in such way that the catliamma originally
represent four broad plates, between which only four narrow radial canals remain over;
these are connected in a supplementary manner by a secondary coronal canal. In the
Acraspedse, on the other hand, which are descended from the Scyphopolyps (with the
four characteristic tamiola), these gastral ridges are the starting-point of the concresence;
at four interradial points the dorsal point of the four tamiola (on the notumbrella) is
fused with their ventral part (on the ccelumbrella) in such a way that the cathamma
originally represent four small nodes or narrow ridges, between which the four broad
radial pouches remain free ; these communicate at the peripheric umbrella margin below
the four nodes, by a primary coronal canal, the distal remains of the simple gastral space.
Although these two essentially different principal forms of the coronal intestine in the
two sections probably express a comprehensive primary difference in its conformation,

REPORT ON THE DEEP-SEA MEDUSAE.

lxxxv

modifications of it are present in both sections, which make it possible to confound the
formations derived from them. The Narcomedusse, especially, approach the Acraspedee
strikingly, the Cubomedusse the Craspedotse. In spite of this, the two groups are
genetically entirely different, and it is therefore quite to the purpose to consider the
two principal forms of the coronal intestine separately ; we designate the former, the
vascular corona, the latter the pouch corona.

§. 122. Vascular corona of the Craspedotse (" corona canalium "). In the majority
of the Craspedotse one and the same form of the coronal intestine is maintained, the
vascular corona, which must be regarded as the typical and original form for this section.

Fig. M. Eucopc campanulata (Leptomeduste, Eucopidaj).

Subumbral aspect, [a) Quadrate oral opening, (s) Ovaries, (qj) Perradial canals. («•) Coronal canals
(v) Velum, (ov) Velar marginal vesicels, (adradial). (h) Umbrella cavity, (re) nerve ring, (oc)
Ocelli at the swollen bases of the tentacles, (ug) Gelatinous substance of the umbrella. {() Tentacles.

It is composed of four narrow, perradial canals (fig. M cp) which run out from the four
angles of the central stomach, and pass through the subumbrella to the umbrella margin,
where they are united by a coronal canal (cc) (comp. Pis. I. II.) The typical quadrUocula r
vascular corona of their Hydropolyps has arisen from the simple gastral space of their
predecessors the Hydropolyps, by their two walls (calyx wall and peristom wall) being
laid together and fused in such a way that only four narrow radial canals remain open
between them. The broad concrescentic surfaces between the four perradial canals still
contain in the developed Craspedotse the important (originally double) layer of

lxxxvi THE VOYAGE OF H.M.S. CHALLENGER.

endoderm cells which authenticate their origin (cathammal plate, § 101). The marginal
coronal canal (cc), which connects the four radial canals at the umbrella margin, does not
appear to be the marginal part of the originally simple gastral space (of the Hydropolyps)
which remains open, but to have arisen from the distal ends of the radial canals which
remain open, being connected by marginal processes (hence the secondary coronal
canal). This typical primitive form of the coronal intestine of the Craspedotse is
subject to varied modifications, of which some (Narcomedusse) are so like the pouch
corona of the Acraspedse that they may be confounded with it (" convergent selection ").
§ 123. Radial canals (" canales radiales," cr). In the majority of Craspedotae we
only find four perradial canals, which open into the coronal canal at the insertion of the

Fig. N. Octorchis germanica (Leptomedusse, Eucopid*).

Profile view, (ug) Gelatinous umbrella, [us) Solid gelatinous peduncle of the stomach, (v) Velum.
(to) Velar marginal vesicles, (tp) Perradial tentacles, (ti) Interradial tentacles. («') Distal testis
(on the subumbrella). (s2) Proximal testis (on the oesophagus), (cp) Perradial canals. (</) Stomach.
(al) Oral lobes.

four primary tentacles. They are usually very narrow and cylindrical, more rarely flattened
like a ribbon (in some Tiaridse, Geryonidse, Narcomedusse). There are rarely six canals
instead of four (in part of the Cannotidse and Geryonidse, System, taf. ix. figs. 6-9 ;
taf. xviii. figs. 7, 8). In some families eight canals are constant, as four interradial
secondary canals are developed supplementarily in the middle between the four perradial
(Pis. III.— VIII. ; Melicertidse, Octocannidse, Trachynemida?, Aglauridse, System, taf. viii.,
xvi., xvii.). On the other hand we rarely find eight adradial canals, which have arisen
from basal bifurcation of the four perradial (Ctenaria, Cladonema, Dendronema, System,
taf. vii.). In the iEquoridse the number of the radial canals rises from 8 to from

REPORT ON THE DEEP-SEA MEDUSAE. lxxxvii

50-60, sometimes to from 100-200 and more, and is at the same time very inconstant
(System, taf. xiv., xv.). Their number is likewise variable and inconstant in the
Narcomedusse, where it never rises beyond 32 and usually amounts to between
11 and 20 (System, taf. xix., xx.). Whilst the radial canals of most Craspedotse, are
quite simple and unbranched, the three sub-families of the Cannotidse, are distinguished
by their being repeatedly ramified ; the radial canals of the Polyorchidse are beset with
csecal side branches, and consequently appear pinnated (Ptychogena, PI. II.) ; in the
Berencidae both the simple or ramified side branches of the radial canal, and their
direct process open into the coronal canal ; in the Williadse the radial canals are
bifurcated as in the Cladonemidse and Zygocannidae ; the bifurcations may also be
repeatedly dichotomised (System, taf. ix.). In those Craspedotse where a solid gela-
tinous gastral peduncle is developed in the centre of the subumbrella, each radial canal
is divided into two sections, an ascending peduncle canal, and a descending subumbral
canal ; the peduncle canal leads from the bottom of the stomach at the oral end of the
gastral peduncle to the base of the latter (in the bottom of the suburnbrella) ; the
subumbral canal from the last-named point to the coronal canal (fig. N).

§ 124. Coronal canal of the Craspedotse (" canalis circularis," cc). In all Craspedotse
the radial canals are originally connected at the umbrella margin by a coronal canal
(" canalis circularis "). Its ontogenesis shows that it does not arise from the primary
gastral space remaining open at its peripheric margin, but by secondary anastomotic
formation of the radial canals. Whilst the latter are formed by the two endodermal
surfaces of the simple gastral space of the polyps becoming fused into four broad interradial
cathammal plates, the connective coronal canals at the distal end of the open radial canals
are formed by the separation of their two epithelial layers (dorsal and ventral endoderm)
at the distal margin of the endodermal cathammal plates. The marginal coronal canal
of the Craspedotse, is therefore formed in the same way as the secondary coronal canal of
the Cycloperise among the Discomedusae, whilst the primary coronal canal (or better,
" coronal sinus") of the other Acraspedae has quite a different formation (comp. § 126,
123). The tentacle canals, which pass into the hollow tentacles, run out in a distal
direction from the simple coronal canal of the Craspedotse. The " centripetal canals "
which run out from the coronal canal towards the middle of the subumbrella, where they
end csecally (§ 135), are sometimes developed in a proximal direction.

§ 125. Festoon canal and radial pouches of the Narcomedusse. The order of the
Narcomedusse is distinguished by a peculiar condition of formation of the coronal intestine,
apparently completely different from that of the other Craspedotse. Broad caecal
pouches apparently proceed from the central stomachs, which are not connected by a
coronal canal, and resemble those of the Typhloperiae among the Discomedusse. More
minute comparative investigation however shows that they cannot properly be compared
with the latter, but that the typical fundamental form of the coronal intestine is the same

lxxxviii THE VOYAGE OF H.M.S. CHALLENGER.

here as in all other Craspedotse, i.e., four narrow perradial canals communicating by a
coronal canal at the umbrella margin. This original formation only undergoes peculiar
modifications, which appear to be influenced pre-eminently by the centripetal migration
of the tentacles on to the exumbrella (comp. System, 1879, taf. xix., xx. pp. 302-306).
Whilst in other Craspedotae the tentacles usually retain their original position on the
umbrella margin, in the Narcomedusse they migrate from there up to the dorsal surface
of the exumbrella, towards the apex, taking with them, from their original insertion, part
of the umbrella margin, whose urticating ring is transformed into a centripetal umbrella
clasp (" peronium "), ( § 68, Pis. IX.-XI V., en). The original marginal coronal canal,
which lies on the inner side of the marginal urticating ring, follows the centripetal
processes of the latter, which forms the peronium in the exumbrella and edges the two
lateral margins of the peronium in the form of a double peronial canal (" canalis
peronialis," ck, Pis. IX. -XIV.). The two parallel double canals (which are only separated
by the peronium) open at the dorsal point of insertion of the tentacle, into the distal end
of the true radial canal, which is expanded like a pouch (Cunantha, Cunina, System,
taf. xix. figs. 1, 3). Whilst many Cunanthidas (the Cunoctanthida3) show this most simple
condition and are therefore connected immediately with certain Trachomedusas
(Geryonida^), in other Cunanthidse (Cunoctonidas) the distal part of each radial canal
bifurcates into two csecal pouches diverging distally, and these internemal lobe pouches
therefore lie in pairs between each two peronia and peronial canals (Cunarcha, PL IX.
figs. 2-4, Cunoctona, System, taf. xx. figs. 1-6). In the closely allied iEgmidse the
simple proximal part of the widened radial canals has undergone retrograde formation or
become merged into the periphery of the stomach, whilst the two lobe pouches which have
arisen from their distal part are developed into independent internemal gastral pouches, open-
ing immediately into the periphery of the stomach (sEginura, Pis. XIII., XIV.; JEgina,
System, taf. xx.). The double peronial canals, which open into the stomach between
each two pair of pouches, seem at first to be simple radial canals, connected by a simple
coronal canal at the umbrella margin ; this apparently simple coronal canal, however,
really consists of four or eight separate marginal canals (" canales marginales," cm); these
are completely separated by the distal ends of the peronia, and each marginal canal, along
with the inverted halves of the two contiguous peronial double canals, forms a horseshoe-
shaped arch canal. An arch canal opens by two separate ostia between each two peronia
into the gastral cavity. The entire coronal canal, when formed as in the iEginidas of four
or eight separate canals, is called the festoon canal (" canalis festivus," cf, Pis. XIII. , XIV.).
The Peganthidse (Pis. X.-XII.) present another modification. The radial canals are
merged in the central stomach or have undergone retrograde formation, so that the festoon
canal (or the modified coronal canal) opens immediately into the stomach and with twice
the number of openings than the number of arches composing the festoon canal ; each arch
opens by two gastral ostia (go). In the family of the Solmaridas the festoon canal is

REPORT ON THE DEEP-SEA MEDUSAE. Ixxxix

completely obliterated ; the radial canals may also be completely obliterated here
(Solmonetidse), whilst in other Solmaridae, the pernemal gastral pouches remain
(Solmissidae), sometimes the internemal gastral pouches (Solmundinae, System, taf. xix.
xx. p. 346).

§ 126. Pouch corona of the Acraspedae (" corona bursarum "). In contrast to the canal
corona of the Craspedotae, an essentially different form must be regarded as the typical and
original arrangement for the section of the Acraspedae, i.e., the pouch corona composed of
four wide, perradial pouches, which begin at the circumference of the central stomach, and
run in the subumbrella towards the umbrella margin where they are united by a coronal
canal. This typical quadripartite pouch corona of the Scyphomedusae, has been developed
from the simple gastral space of their ancestors the Scyphopolyps by the four interradial
taeniola of the latter being laid together and fused at four points (of equal height), or in
four streaks, by their upper dorsal parts and lower ventral parts. In this way four
small interradial nodes or narrow ridges are originated (PI. XV. figs. 2, 3, hi ; PI. XVI.
figs. 2, 3, ks), which form incomplete septa between four wide perradial pouches {bp).
The small fused nodes or the four narrow fused ridges in the mature Acraspedae still
contain a double layer (or a layer become simple by fusion) of endoderm cells, which
indicates this origin as acathammal plate (§ 101, PI. XXV. fig. 8). The marginal coronal
canal which connects the four broad pouches at the umbrella margin below the four
narrow septa, appears to be the marginal part of the originally simple gastral space of the
Scyphopolyps, which has remained open (therefore a " primary coronal canal "). This
typical primitive form of the coronal intestine of the Acraspedse is subject to varied
modifications, of which some (Flosculidae) are so bike the canal corona of the Craspedotas
that they may be confounded with it (" convergent selection ").

§ 127. Eadial pouches (" bursas radiales," bs). The constant number of four of the
perradial pouches of the Acraspedas is very remarkable when contrasted with the frequent
variations of the number of four shown by the Craspedotae. The four perradial pouches
are orignally (phylogenetically) present, though they are often lost at an early stage.
They correspond to the four flat gastral pouches of the Scyphostoma, which are constantly
separated by four interradial taeniola. Frequent individual exceptions (especially
individuals with six or eight radial pouches) are of no importance, as they do not
transmit their peculiarity. The two sub-sections of the Acraspedaa, the Tesseronise and
Ephyronise, are, however, in general strikingly distinguished, as in the Tesseronise (Pis.
XV., XXVI. ), the four primary radial pouches always remain very large, and with the
coronal sinus belonging to them form the permanent principal component part of the
coronal intestine, whilst in the Ephyroniae (Pis. XXVII. -XXXII.) they have usually
undergone retrograde formation or become merged by fusion into the central stomach ;
they appear, however, to be of no great importance, and to have retreated towards the
peripheric pouch corona. The latter, which consists of at least eight, but usually of

(ZOOL. CHALL. EXP. — PAET XII. — 1881.) M m

XC THE VOYAGE OF H.M.S. CHALLENGER.

sixteen radial coronal pouches, always forms the principal component part of the coronal
intestine in the Ephyroniae, whilst in the Tesseroniae it has retreated completely against
the inner quadrilocular pouch corona. Important differences, are, however, connected with
the further organisation of the two sub-sections. It is therefore practical to consider the
radial pouches of the Tesseroniae and Ephyroniae separately.

§ 128. The four perradial pouches of the Tesseroniae. In all the three orders of the
Tesseroniae, the Stauromedusae, Peromedusae, and Cubomedusae, the four primary perradial
pouches (bj)) form, from their circumference and extent, by far the most important part
of the coronal intestine, whilst, on the other hand, the marginal pouch corona has retired
against its periphery (System, pp. 363-449, taf. xxi.-xxvi.). As in all Tesseroniae,
the umbrella is highly vaulted, and its fundamental form represents a high quadrate
pyramid (usually truncated above), the four broad pouches occupy its four lateral
surfaces, whilst the intermediate four interradial septa correspond to the four angles of
the regular pyramid. Each quadrangular pouch communicates at the upper or proximal
margin by a fissure-shaped gastral ostium (go) with the central stomach, at the lower or
distal margin by two or more fissures with the marginal pouches ; its two lateral
margins are formed by the septa or cathamma, and their ideal (interradial) prolonga-
tions. All the four perradial pouches communicate below the four cathamma, by four
interradial fissures, which taken together represent an ideal primary coronal canal,
the coronal sinus ("sinus coronaris," § 134). Of the two flat, quadrangular walls
of each perradial pouch, the outer (abaxial) is formed by the inner (concave) endodermal
surface of the notumbrella, whilst the inner (axial) is formed by the endodermal surface of
the coelumbrella. Whilst all Tesseroniae agree without exception in these essential
general conditions of structure, many important modifications are found in detail which
are referable on the one hand to the different extent of the cathamma, and on the other
to the different position of the gastral ostia. In one portion of the Stauromedusae, and
in all Peromedusse, the gastral ostia form narrow longitudinal fissures of the central
stomach, standing more or less vertically, whilst in another portion of the Stauroniedusae,
and in all Cubomedusae, they form broad transverse fissures, standing more or less
horizontally ; the special anatomical relation of the proximal part of the pouch to the
central stomach, therefore, differs considerably. As regards the four interradial septa of
the four pouches or the four primary cathamma, in one portion of the Stauromedusae
(Tesseridae), and in all Peromedusse, they form small, but very firm nodes, as hard as
cartilage (" nodi cathammales," PI. XV. figs. 2-6, hi; Pis. XX. -XXIV., hn) ; whilst in
the other portion of the Stauromedusae (Lucernaridae), and in all Cubomedusae, they form
long, narrow ridges (" limites cathammales," Pis. XVI., XVIL, hs). The Peromedusae are,
moreover, distinguished from all other Tesseroniae by the union of the four perradial
pouches above the four septal nodes, into a powerful upper coronal sinus, which
encircles the central stomach, and reaches with its upper (closed) margin up to the
pyloric ring (Periphyllidte, Pis. XX., XXL, co" ; comp. § 134).

REPORT ON THE DEEP-SEA MEDUSA. XCl

§ 129. The four perradial pouches of the Ephyroniae. In the second and younger
sub-section of the Acraspedae, the Ephyroniae, the coronal intestine strikes out a direction
of formation which contrasts thoroughly with, and differs essentially from that presented
by the first and older sub-section, the Tesseroniae. Whilst the four primary radial
pouches of the coronal intestine in the Tesseroniae are always very large, and form its
principal component part, in the Ephyronias they appear to be wanting, or only to exist
as small rudiments : up till now they have been completely overlooked in the Ephyroniae
or Discomedusae, and not taken into account by any author on Medusae. The two deep-
sea Cannostomae, Nauphanta (Pis. XXVIL, XXVIII.) and Atolla (PL XXIX.), are of
great importance for comprehending their formation ; from their size they show the
primitive formation more clearly than the known small Nausithoe, in which the four
septal nodes are certainly also present but are very small, and have hitherto been
invariably overlooked. In all the Cannostomae named (and probably in all Discomedusae
of this sub-order) four interradial septal nodes (hi, PI. XXVIL fig. 3 ; PL XXVIII. figs.
14, 15 ; PL XXIX. figs. 3, 6) are found in the upper part of the coronal intestine. They
correspond completely, in situation and significance, to those of the Tesseridae (PL XV.
figs. 2-6, Jen), and the Periphyllidae (Pis. XX.-XXIV., hi). In Atolla (PL XXIX.) these
important four interradial cathammal nodes are triangular and strongly flattened, whilst
in Nauphanta (Pis. XXVIL, XXVIIL), and also in Nausithoe, they are very small, and
have as yet been taken no notice of. In their transverse sections, however, they show
distinctly the important cathammal plate or endodermal lamella ()'k), the fused plate
between the gelatinous body of the notumbrella and the supporting plate of the
ccelumbrella (PL XXV. fig. 8, dk). The ventral wall of the coelumbrella and the dorsal
wall of the notumbrella are fused by means of these four interradial fused nodes ;
between these four perradial transverse fissures remain, which represent the four
rudimentary, very much shortened radial pouches, and whose proximal margin is at
the same time to be regarded as a "gastral ostium" (go). If we suppose these nodes
prolonged interradially downwards in the form of narrow septal ridges, the underlying
coronal canal or coronal sinus (cs) will be thereby divided into four broad, long,
radial pouches, resembling those of the Lucernaridae (Pis. XVI., XVII. , bp). In the two
other sub-orders of the Discomedusae, the Semostomae and Rhizostomae, the four primary
septal nodes, which were originally present, appear to have undergone retrograde
formation, and become lost, whilst the Cannostomae have kept them faithfully up to the
present day ; the Cannostomae — which in many respects still resemble the Tesseroniae —
being the ancestral form of the Discomedusae, from which the Semostomas have been
developed later (and the Rhizostomae still later from the Semostomaa). In the
Ephyrula larva of the latter, the four primary gastral filaments, probably indicate the
spot formerly occupied by the predecessors of the septal nodes. The four cruciate
pouches of the central stomach already described, may perhaps be considered as partly

xcii THE VOYAGE OF H.M.S. CHALLENGER.

or wholly the remains of the original four radial pouches of the coronal intestine, and the
centripetal septal ridges between them, as remains of the cathammal plates.

§ 130. The marginal pouches of the Tesseroniae. It is only in the Tesseridae, the first
and oldest family of the Tesseroniae, that the peripheric coronal intestine is formed
exclusively of the four large perradial pouches and the marginal coronal canal or
coronal sinus, which connects the latter below the four interradial septa (Tesserantha,
PI. XV. figs. 2-6 ; Tessera, System, PL XXL). In all other Tesseroniae, other pouches
are found at the distal margin of the four perradial pouches, which are developed as
peripheric archings outwards of the latter and compose the corona of marginal pouches.
In the Lucernaridaa they are represented by four pair of arm pouches which pass into
the eight adradial arms or marginal lobes, from whose distal end they send out a tentacle
canal into each tentacle (Pis. XVI., XVII. ). The Charybdeidae, in which the eight
adradial marginal pouches are separated alternately by the four interradial septa and the
four perradial " frenula velarii " (PL XXVI. ), comport themselves quite in the same way.
In the Chirodropidae the number of radial pouches is doubled, as each of the eight
adradial marginal pouches is split up into two subradial lobe pouches (System, p. 446,
taf. xxvi.). In most Cubomedusae, moreover (with the exception of the lowest forms,
the Procharagmidaj) numerous simple or branched velar canals run out from the
distal margin of the lobe pouches, extend in the velarium (the broad marginal membrane
connecting the marginal lobes) and end caeeally with anastomosis (PL XXVI. fig. 8, cv).
The conditions differ somewhat in the Peromedusae, as in the Pericolpidae, first of all
(System, taf. xxiii.) eight principal coronal pouches (four perradial and four interradial)
run out from the distal margin of the coronal sinus. Their subumbral walls form the
eight coronal plates of the coronal muscle. Each of the eight coronal pouches is split up
into a medial principal canal, and two distal lobe canals. The medial canal passes into the
tentacle in the four perradial coronal pouches, into the sense clubs in the four interradial ;
whdst, on the other hand, the two lobe canals of each coronal pouch (bl) provide for the
halves of each two adjacent canals turned to each other, at whose distal end they
communicate by a horseshoe-shaped canal (bw) with the opposite canals of the other halves
(PL XXII. fig. 22 ; PI, XXIII. fig. 29). The Periphyllida? are only distinguished from
the Pericolpidas by having three tentacles with two intercalary subradial marginal lobes
in the place of each simple perradial tentacle (Pis. XVIII.-XXV). As the whole lobe
canals of the Peromedusa are connected below by U-shaped horseshoe canals (bn) at the
distal end of the marginal lobes, a waved connected festoon canal of pecidiar formation
is developed in all Peromedusae at the outermost margin of the lobed umbrella (PI. XX.
figs. 8, 22).

§ 131. The marginal pouches of the Ephyronia?. Whilst in the Tesseronia? the
marginal pouches always appear as merely subordinate appendages on the distal margin
of the four perradial pouches (or of the large coronal sinus formed by junction of the

REPORT ON THE DEEP-SEA MEDUSAE. xciii

latter), in the Ephyroniae or Discomedusae they form the principal component part of the
coronal intestine. They differentiate here in a great variety of ways. The more the four
primary perradial pouches in the Discomeclusse become subsidary, the more extensive are
the numerous and voluminous marginal pouches which take their place ; in spite of this,
the latter must be regarded here as originally distal processes of the former. It is
important to distinguish the original from the later formations among the manifold and
protean differentiations undergone by the corona of marginal pouches in the different
groups of Discomedusaa. We consider the simple eight principal radial pouches (four
perradial and four interradial) which appear universally in the Ephyra larva of the
Ephyronise, and then pass into the eight typical sense clubs or rhopalia (and which are
therefore called sense pouches, ocular pouches, or rhopalar pouches (" bursas rhopalares," br)
as the primitive marginal pouches of this order. They correspond to the eight coronal
pouches of the Pericolpidae (among the Peromedusse), of which four likewise lie perradially
and four interradially, the former passing into the tentacles, the latter into the rhopalia
(woodcut, fig. 0, p. xcvi.). Eight adradial alternating tentacle pouches, which pass into the
eight typical tentacles of this order (Pis. XXVII., XXVIII., bf), are, however, immediately
added to these eight principal rhopalar pouches. The characteristic corona of marginal
pouches in most Ephyronige, therefore consists of sixteen coronal pouches ("bursas coro-
nares," be), of which the eight principal rhopalar pouches are, however, phylogenetically
older than the eight adradial tentacle pouches. All the sixteen pouches are separated by
sixteen narrow septa or fused streaks ; these sixteen subradial cathammal ridges correspond
to the sixteen subradial peronia of the Periphyllidas. The sixteen coronal pouches of the
Discomedusse do not, however, communicate like those of the Periphyllidae by a marginal
festoon canal (below the distal ends of the cathammal ridges), but remain originally
completely separate and end csecally. Many Ephyridse still show this original condition,
as the eight large or principal pouches end csecally in the eight sense clubs, the eight
smaller adradial pouches at the bases of the eight tentacles. The varied peripheric
pouch formations of the Discomedusse may all be derived from this primary condition as
secondary modifications and more recent developments.

§ 132. Typhloperiae, or Ephyronias without annular canal. According to the preceding
view, the peripheric pouch corona of the DiscornedusaB consists originally in all Acraspedae
of this order (as in all their Ephyrula larvae at the present time) of sixteen simple, caecal
coronal pouches, separated by sixteen narrow subradial cathammal ridges, and from which
the eight principal pouches (four perradial, four interradial) pass into the rhopaba, and
the eight adradial into the tentacles. As regards the further differentiation of this
originally simple pouch corona, there are two different types in this order so rich in
forms ; the older, conservative type, we shall term for brevity, the Typhloperiae, the
young and more progressive type the Cycloperiae ; the latter is distinguished by the
accprisition of a second circular canal, which is still wanting in the former. Of the

XC1V THE VOYAGE OF H.M.S. CHALLENGER.

Discomedusse, the group of the Typhloperise, or Ephyronise, without coronal canal, includes
first all the Cannostpmse (Ephyridse, Linergidse) and, secondly, half of the Semostomse
(Pelagidse, Cyaneidas). In these the sixteen coronal pouches always continue completely
isolated, whether they remain simple or become repeatedly branched. The tentacles
remain simple and unbranched in the Ephyridse (System, taf. xxvii., xxviii.) and Pela-
gidse (System, taf. xxxi.) each coronal pouch is, however, usually bifurcated into two lobe
pouches, which provide for the inverted halves of each two neighbouring canals ; the two
simple csecal lobe canals of each lobe therefore belong to two different adjacent marginal
lobes and remain completely separated by the (subradial) peronium. Consequently,
there are usually thirty-two csecal lobe pouches present, but this number may be
increased considerably by later division and the formation of secondary lobes. The
Lineragidse (System, taf. xxix.) and Cyaneidse (System, taf. xxx.), on the other hand,
are distinguished by distal ramification of the lobe pouches. Each lobe pouch sends out
numerous bifurcate or dendritic canals from their periphery ; these comport themselves like
the " dendritic velar canals " of the Cuboinedusse (PL XXVI. ), never form anastomoses,
and end everywhere with csecal branches at the periphery of the lobes (Pis. XXX., XXXL).
§ 133. Cycloperise, or Ephyronise with annular canal. Whilst in the Typhloperise,
described above, the sixteen original coronal pouches and their distal ramifications remain
entirely separate, the second half of the Discomedusse is distinguished by the development
of a connecting coronal canal between the coronal pouches. Of the three sub-orders of
the Discomedusse, the Cycloperise or Ephyronise, with coronal canal, includes, first the
half of the Semostoruse (Flosculidse, Ulmaridae), and, secondly, the whole of the
Rhizostomse. It is clear from the ontogeny of the Cycloperise that they are derived
phylogenetically from the Typhloperise, and that their characteristic annular canal (" canalis
annularis," ck) has been developed secondarily, as a circular, peripheric anastomosis
between the originally separate, csecal coronal pouches. The Ephyrula larva of the
Cycloperise is so like that of the Typhloperise that they may be confounded together, and
is also furnished with sixteen simple, csecal, coronal pouches. In the course of their
metamorphosis, these pouches in the Cycloperise form lateral processes which grow towards
those of the neighbouring coronal pouches with which they enter into open anastomosis.
In this way the secondary annular canal of the Cycloperise (" canalis annularis ") is
formed ; it is analogous to, (not homologous with) the secondary circular canal (" canalis
circularis ") of the Craspedotse, but different from the coronal canal or primary ring-
canal of the Acraspedse (" canalis coronaris "). The latter lies at the proximal side of the
coronal pouches (between them and the four septal nodes), whilst the former Hes on the
distal side of the coronal pouches. Another important peculiarity of the Cycloperise
stands in direct correlation to the development of their annular canals ; the sixteen broad
coronal pouches of the Typhloperise are transformed in all Cycloperise into sixteen narrow
coronal canals, as the sixteen narrow subradial septal ridges between the pouches

REPORT ON THE DEEP SEA MEDUSAE. XCV

(" limites cathammales ") are extended into sixteen broad septal plates (" tabulae
cathammales "). This originates a special form of the coronal intestine, which resembles
the common form of the Craspedotse so strongly that the two may be confounded, but
which has arisen quite independently, and also differs essentially in its relation to the
principal intestine. Moreover, the narrow " radial canals " of the Craspedotse spring
immediately from the central stomach, whilst the similar "coronal canals" of the
Cycloperise spring originally from the distal margin of a coronal sinus, lying at the distal
side of four cathammal nodes. Otherwise, both the further differentiation of the radial
canals and their circular canals and their relation to the adjacent orders of the umbrella
margin show the greatest similarity in both sections. The oldest family among the
Cycloperise, the Flosculidae (System, taf. xxxii.) show simple radial canals which are not
branched inside the annular canal (as in the majority of the Craspedotse). All the
remaining Cycloperise (the Ulmaridse and all Rhizostomse) have branched radial canals,
which branch inside the annular canal (as in the Cannotidse among the Craspedotse).
In both cases peripheric canals of the annular canal run from it, outside it, into the
tentacles, rhopaba, marginal lobes, &c. (§ 135).

§134. Coronal sinus of the Acraspedse (" sinus coronaris, canalis coronaris "). The
coronal sinus of the Acraspedse (cs) already mentioned forms a very important arrange-
ment, which has hitherto never been properly taken into account. It must not be
confounded with the annular canal described above, but must be considered as an original
and typical peculiarity of this section. Originally, it is a perfectly simple, cybndrical
or flattened ring in the coronal part of the subumbrella, and is formed by the entire
peripheric section of the coronal intestine, lying outside the distal margin of the four
interradial cathammal nodes (hi) or ridges (Jcs), as in Tesserantha (PL XV. figs. 2-6, cs)
and in Tessera (System, taf. xxi., cs), the closely allied ancestral form of all Acraspedse.
Although its ontogeny is still unknown, we may assume with certainty as regards its
phvlogeny, that it does not represent a secondary marginal communication of the four
broad perradial pouches, but rather the marginal portions of the originally simple
Scyphostoma stomach which has remained open, and lies outside the four interradial
cathamma (k). As the latter are originally merely small fused nodes (as in the Tesseridse
and Peromedusse), the intermediate perradial pouches at first merely form narrow
horizontal fissures. As soon, however, as the nodes become elongated concrescentic
ridges (as in the Lucernarida? and Cubomedusse) the insignificant horizontal fissures
extend simultaneously into pouches of considerable size, at the cost of the broad coronal
sinus, which is for the most part absorbed into them. The broad coronal sinus is thus
reduced into a narrow coronal canal which keeps up the communication between the
four large perradial pouches below the distal margin of their septal ridges (PI. XVI. figs.
2, 3, 12. cc ; PI. XXVL). The coronal sinus is most widely extended, and at the same
time undergoes a very striking modification in the Perornedusre, probably in direct

XCV1

THE VOYAGE OF H.M.S. CHALLENGER.

correlation with the powerful extension of the four subumbral funnel cavities (ci) which
here grow centripetally up as far as the pylorus (gy)- The coronal sinus extends corres-
pondingly upwards to the pylorus, and is divided into an upper and a lower part. The
upper coronal sinus (fig. 0, cs') lies above the four small interradial septal nodes (hi),
and may probably be really considered as a peripheric section of the central stomach (gc),
whilst the lower coronal sinus (fig. 0, cs) lies below the septal nodes and probably alone

Fig. 0. Pericopla quudrigala (Peromedusa?, Pericolpa).

Interradial section, (ug) Gelatinous umbrella, (a) Coronal sinus (cs') pro'ximal part, (cs2) distal part,
(s) Genitalia. {Jen) Catliammal nodes, {go) Gastral openings, {md) Deltoid muscles, {mc)
Coronal muscle, {ft.) Trcniola. (gb) Niches of the basal stomach between the tamiola. {gy)
Pylorus, (</c) Central stomach, {gp) Palatine opening, (go) Buccal stomach (oesophagus), {am)
Oral margin, (oi) Sense clubs (interradial). {oa) Ampulla? at their base. (6m) Horseshoe-shaped
canal of the marginal lobes {la), {kl) Peronia between the two limbs of the canal {bl). {tp)
Tentacles (perradial).

corresponds to the true primary coronal canal of the other Tesseronise. If this view be
correct, the true gastral ostia (and at the same time the four perradial pouches) are repre-
sented by the horizontal cleft spaces between the four nodes, whilst the vertical fissures in
the wall of the central stomach, which we have described as gastral ostia (fig. 0, go) must
be regarded as four cruciate pouches of the central stomach. It is possible, however, that
the upper coronal sinus (cs') has arisen from the secondary confluence of the four large

REPORT ON" THE DEEP-SEA MEDUSAE. XCvii

perradial pouches, of whose long septa the distal part only remains (in the form of four
small nodes). Among the Ephyroniae (or Discomedusas), the Cannostomaa only show
similar conditions ; Nausithoe and Ncmphanta (Pis. XXVIL, XXVIII.), and also
Collaspis and Atolla (PI. XXIX.) have preserved the coronal sinus (cs) along with the
four primary cathamma (k) and their four intermediate perradial fissures up to the
present time. In the other Ephyroniae (Seniostomae and Khizostomae), the four
cathamma are dissolved and have disappeared ; the coronal sinus is consequently merged
in the central stomach of which it forms the peripheric part.

§ 135. Peripheric processes of the gastrovascular system. In many lower Medusae of
both sections, the formation of the gastrovascular system is limited to the essential portions
of the principal intestine and the peripheric coronal intestine already described. In the
majority of Medusae, however, there are additional, accessory processes of the vascular
system, which serve for the nutrition of the peripheric organs. Such pre-eminently are
the nutritive canals of the marginal organs, the tentacles and the organs of sense : they run
out partly immediately from the distal ends of the radial canals or pouches, partly from
the marginal coronal canal, connecting the latter among the Craspedotae; besides these,
special centripetal canals are sometimes developed, which grow out from the annular canal
towards the middle of the subumbrella, and serve for the nutrition of the latter. They are
found almost exclusively in the order of the Trachomedusae, and are usually simple with
caecal proximal ends : e.g., as in Pedis among the Pectyllidae (PI. V. fig. 2 ; PL VI. fio-s.
11, 20), in Olindias among the Petasidse, and in part of the Geryonida? (System, taf. xv.
figs. 9, 10 ; taf. xviii. figs. 5, 8). The Cannotid Spirocodon is distinguished by dendritic
ramification of the centripetal canals (System, p. 636, No. 588). The most remarkable of
these peripheric processes of the vascular system among the Acraspedae are those which
pass into the marginal lobes or the velarium formed by their fusion. The velar canals of
the Cubomedusae (PI. XXVI. fig. 8), the lobe canals of the Discomedusa? belong to this
category. In the Typhloperiae (§ 132), as in the Cubomedusaa, the latter never form
anastomoses, whilst in the Cycloperiae (§ 133) they form by repeated anastomoses a
vascular network, varied in form, which often occupies not only the marginal lobes but a
large part of the subumbrella (PI. XXXII. fig. B, §117). Finally, there are noticeable
irregular gelatinous canals, sometimes simple, sometimes branched (" canales endo-
collares ") which in the larger and older Discomedusae grow out extensively from the
coronal intestine into the gelatinous substance of the body (e.g., Chrysaora, Cyanea,
Pilema, Crambessa).

§ 136. Peripheric openings of the gastrovascular system. In the majority of all
Medusae the central mouth forms the only opening of the gastrovascular system. In
single Medusae of both sections, however, there are also small peripheric openings, which
are in every case of a secondary nature. The most important of these are found on the
umbrella margin, and may be generally termed marginal pores (" pori marginales ").

(ZOOL. CHALL. EXP. PAKT XII. — 1881.) M n

XCVUl THE VOYAGE OF H.M.S. CHALLENGER.

Among the Craspedotae the so-called excretory papillae or subumbral papillae (also termed
marginal funnels or excretory funnels, "papillae excretoriae" or "subumbrales") belong to this
category. These peculiar excretory organs of the umbrella margin are found exclusively
in the order of the Leptomedusae, among which, however, they are tolerably wide spread.
They are small funnel-shaped or conical warts, which project in different numbers, on the
distal margin of the subumbrella between the margin of insertion of the velum and the
circular canal : they contain an evagination of the circular canal and open by a small
aperture into the umbrella cavity (System, p. 119, taf. xi. fig. 13, ex; taf. xiii. fig. 5, q).
Their number in many Eucopidae (e.g., Octorchis) and iEquoridae (e.g., Polycanna) is very
considerable but indefinite. The flagellate cells of the endoderm in the subumbral papillae
are directed externally towards the marginal pore, so that the latter must be regarded
as an "excretory opening" or "anus." Among the Acraspedae, similar marginal axial
openings have been long known in Aurelia, where they lie at the distal end of the eight
adradial canals, where the latter open into the coronal canal ; the ejection of fluid by these
canals can be easily observed directly in the eight adradial marginal pores. These pores
also occur in other Ulmaridae. As, however, they are small and easily overlooked, it is
possible that they are much more widespread than we are aware of at present. In some
Medusae, the tentacles seem also to have an opening at the distal end.

§ 137. Reproductive organs (" gonades, genitalia, sexualia," s). In all Medusae the
reproductive organs show very simple, homologous, and uniform conditions of formation,
inasmuch as they everywhere consist essentially of reproductive glands (" gonades "), and
are universally developed in the subumbral wall of the gastrovascular system. The two
sexes show no essential difference, as the male spermaria develop in the same places, and
in the same way as the female ovaries. On the other hand, there is an essential and
thorough distinction between the two sections of the class Medusae, inasmuch as the
subepithelial layer of cells, which, as sexual epithelium or germinal epithebum, furnishes
the two kinds of sexual cells, the spermatozoa, and the ova, belongs to the ectoderm in the
Craspedotae, to the endoderm in the Acraspedae. In the former, the mature reproductive
elements are therefore emptied immediately outside into the ectodermal umbrella cavity,
whilst in the latter they first pass into the endodermal hollow space of the gastrovascular
system, from which they are ejected by the mouth. The Craspedotae are therefore
Ectocarpae, like the Hydropolyps, Siphonophorae, and Ctenophorae, whilst the Acraspedae
are Endocarpae, like the Scyphopolyps and Corals (§ 19).

§ 138. Gonochorism and Hermaphroditism. Nearly all known Medusae are of separate
sexes, gonochoristic, only a very few are hermaphrodite. The Pelagic Chrysaora certainly
belongs to the latter (System, p. 503, taf. xxxi.). Here, whilst the four interradial
genitalia produce ova, roundish spermaria or testes-like sacs are simultaneously formed in
very irregular number on the most varied parts of the sumbumbral wall of the gastro-
vascular system, both in the genitalia, and on the oral arms, and also on different parts of

REPORT ON THE DEEP-SEA MEDUSAE. xcix

the principal intestine, and of the coronal intestine, and even in the peripheric coronal
pouches. Chrysaora is usually purely male in youth, hermaphrodite later on, and purely
female when mature ; it seems, however, also to furnish purely gonochoristic specimens,
which during their whole lifetime form purely male or purely female sexual cells.
Hermaphroditism simUar to that of Chrysaora seems also to exist in the closely allied
Linergidas ; the peculiar, regularly distributed subumbral vesicles of these CannostomaB
appear to develop spermatozoa, whilst their genitalia only produce ova (System, p. 493, taf.
xxix.). Among the Craspedotse part of the Narcomedusse appear to be hermaphrodite ;
these and some other (probable) cases of hermaphroditism are not known with sufficient
certainty. The sperm cells of the Medusae are universally fine flagellate cells, not strik-
ingly distinguishable from those of other Acalephse. The egg cells are usually naked and
amoeboid in the Craspedotse (PI. I. fig. 8) but usually enclosed in fulcral capsules in the
Acraspedse (PL XX. fig. 7) ; in some Acraspedse they contain a visible food-yolk and
then come to more than 1 mm. in size (PI. XXV. fig. 4).

§ 139. Genitalia of the Craspedotse. The reproductive glands are developed in all
Craspedotse (in thorough and important contrast to the Acraspedse) from the ectoderm of
the subumbral wall of the gastrovascular system, and when mature, are emptied
immediately into the umbrella cavity. They are sometimes formed more in the central
part of the gastrovascular system, sometimes more in the peripheric part ; in the first
case we call them gastral genitalia, in the second case, vascular genitalia. Gastral
genitalia in the subumbral wall of the central stomach, and the oesophagus proceeding
from it, are found in the two orders of the Anthomedusae (PI. I.), and the Narcomedusse
(Pis. X.-XIL), whilst on the other hand, the two orders of the Leptomedusse (PL II.), and
the Trachomedusse (Pis. III. -VIII.) have vascular genitalia in the subumbral walls of the
peripheric radial canals. There are, however, isolated exceptions in both groups ; the
central gastral genitalia sometimes grow centripetally and extend from the stomach also
on to the peripheric canals (e.g., Nemopsis among the Anthornedusae, System, taf. v.
figs. 6-9) ; in many they even pass mostly or entirely on to the gastral pouches of this
order, which originate from the proximal part of the radial canals (Pis. IX., XIII., XIV.).
In other cases the reverse occurs, and the peripheric vascular genitalia extend upon the
central stomach (e.g., Staurostoma, Staurophora, Orchistoma among the Leptomedusse,
System, taf. viii. fig. 6 ; taf. xv. figs. 3-5). If we consider the Craspedotse to be a
monophyletic group of animals, the gastral production of the genitalia may probably be
regarded in general as the older and original condition from which the vascular
production of the reproductive glands has been developed later, or we may assume that
the subumbral wall of the entire gastrovascular system originally produced sexual cells,
and that these productions were distributed later on the principal intestine and coronal
intestine. If on the contrary (and with more probability) we consider the Craspedotse
as a polyphyletic group, the vascular genitalia and the gastral genitalia may have arisen
in the different orders, independently of one another.

C THE VOYAGE OF H.M.S. CHALLENGER.

§ 140. Gastral genitalia of the Craspedotae. The simplest and most primitive
condition of the genital formation is shown in the section of the Craspedotae in such
genera as have only a simple reproductive organ, a circular genitalium in the subumbral
gastral wall, having the mouth placed in the middle. This condition is shown by the
Codonidae among the Anthomedusae, and also by isolated genera among the Anthomedusae.
The Codonidae (System, p. 10, taf. i., ii.) are distinguished by a long thin oesophagus,
extending in a vertical direction, in whose wall the reproductive cells are regularly
developed, so that the genital ring keeps the form of a cylindrical tube. In contrast to
this formation the Narcomedusse have a flat, broad, gastral sac, extending in a horizontal
direction, so that the genital ring in its subumbral wall likewise appears flat and broad
(Poli/colpa, PI. X. fig. 1). Whilst in many Narcomedusae, the central genital ring
remains quite simple, in others, it extends centripetally on to the peripheric radial
pouches, and in many it is finally limited to the radial pouches. It is divided into the
same number or double the number of separate genital pouches, which originally appeared
as radial canals (comp. the System der Medusen, pp. 312, 327, 335, 347, taf. xix., xx.).
In JEginura (Pis. XIII., XIV.), the genitalia are formed by sixteen internemal gastral
pouches, in Pegantlia (PL XIII. fig. 3) by a corona of separate sacs, which evaginate
separately into the periphery of the stomach and hang down in the separate subumbral
lobe cavities of the umbrella corona (PI. XL fig. 3). In most Anthomedusse the
originally simple genital ring (in the Codonidae) is divided in the same way, in four or
eight radial pieces. The radial division first takes place by the four interradial areae
remaining free from sexual production, whilst four perradial genitalia lie in the four
angles of the stomach (System, taf. iii. figs. 1,2; taf. iv. 1, &c). Each of these may,
however, be re-divided into two halves, which are separated by the perradial longitudinal
muscle of the angles of the stomach (Thamnostylus, PL I. ; System, taf. iv., figs. 3, 10).
Finally, these eight adradial reproductive glands may be placed near each other and
fused in the four interradia in such a way that the inverted limbs of each two adjacent
genitalia, originally separate, are connected into a single gland ; so that, finally, there are
four interradial genitalia in the side walls of the stomach (System, taf. v. figs. 1,3; taf. vi.
fig. 3, 15, &c). We have, therefore, apparently the same condition as in many Acraspedae ;
only in the latter, the interradial position of the four bow-shaped genitalia is a primary
appearance, whilst in the Craspedotae it is secondary (or rather tertiary).

§ 141. Vascular genitalia of the Craspedotae. In antithesis to the gastral genitalia of
the Anthomedusse and Narcomedusae, we find in the Leptomedusae and Trachomedusae
the reproductive glands are usually limited to the subumbral wall of the radial canals,
of which they sometimes occupy the entire length, sometimes only a part of it (proximal,
middle or distal part). As in most Craspedotae the number of the radial canals amounts
to four, they are usually also four perradial genitalia (PL II.). This primitive number is
doubled in those Craspedotae in which four secondary interradial canals have arisen

REPORT ON THE DEEP-SEA MEDUSA. ci

between the four primary perradial (Melicertidse, Octocannidaa, Aglauridae, Pcctyllidaj,
Pis. III. -VIII. ; System, taf. viii. fig. 10; taf. xvi, xvii.). In the iEquoridae, where
the number of the radial canals multiplies indefinitely (20-80 and over), the number
of the genitalia fastened to them increases correspondingly (System, taf. xiv., xv.).
In the simplest case, each reproductive gland forms a simple, ridge-shaped thickening in
the suburethral wall of each radial canal. In many Craspedotae, however, this ridge is
divided into two lateral halves, between which the median line of each wall remains free
from the sexual products. The two paired halves of these reproductive ridges lie further
asunder in proportion to the developement of a longitudinal radial muscle in this median
line (PI. VIII. fig. 9). They sometimes hang down in the umbrella cavity as two
parallel folds (e.g., iEquoridaa), sometimes extend like leaves in the surface of the
subumbral wall (Geryonidae). The canal usually forms a caecal evagination into the
genital ridge on the subumbral wall of the radial canal, whether it remains simple, or is
divided into two paired halves, so that it assumes the form of a pendant vesicle, sac, or
tube (Pis. III.-VIIL). Further complications in its formation also arise when the
genital band is arranged in folds ; the depressions between the separate folds may also
become so deep that it is divided into numerous separate vesicles (e.g., Olindias, System,
taf. xv. fig. 11). The Octorchidse, a sub-family of the Eucopidaa, are distinguished by
each of the four genitalia, being divided into two pieces lying far apart ; the proximal
part lies on the ascending radial canal of the umbrella peduncle (fig. N, s2), the distal
part on the descending radial canal of the subumbrella (fig. N s', p. 190); here eight
genitalia also come on four radial canals.

§ 142. Genitalia of the Acraspedse. The reproductive glands in all Acraspedse (in

constant and typical contrast to the Craspedotae) are developed from the endoderm of

the gastrovascular system. The sexual products in process of development first

appear from the subepithelial germinal layer, in the depth of the gelatinous mesoderm,

and are there enclosed in special fulcral capsules (PI. XXV. fig. 7, zz). When mature,

they break through these chorion sheaths, and fall freely into the gastral space from

which they are emptied out through the oral opening. In the Acraspedse, as in the

Craspedotse, the formation of the reproductive cells takes place sometimes more in the

central part of the gastrovascular system, sometimes more in the peripheric part. Whilst

in the Craspedotse, the central gastral genitalia appear to be the primary formation, and

the peripheric vascular genitalia to be derived secondarily from them, in the Acraspedse,

the development of formation is probably reversed. It is precisely the older and

more simply constructed Tesseroniae (the Stauromedusse, Peromedusae, and Cubomedusae),

that have bursal genitalia, which are developed in the subumbral wall of the four wide

radial pouches, or of the wide peripheric coronal connecting them ; in the younger and

more highly developed Ephyroniae (the Discomedusse), the genitalia pass more and more

centripetally from the subumbral wall of the radial pouches or peripheric coronal sinus,

Cll

THE VOYAGE OF H.M.S. CHALLENGER.

or to the subumbral wall of the broad, flat central stomach, so that they here appear
again as gastral genitalia. Another still more important distinction between the two
sections consists in the fact that in the Craspedotae the four radial pieces of the
reproductive apparatus lie originally perradially, whilst on the contrary, in the Acraspedae
they lie originally interradially. Whilst in the Ephyronige, among the Acrasj:>eda3, there
are usually four interradial glands (more rarely divided into eight pieces), in the
Tesseroniae, on the contrary, there are usually eight separate pieces, which, however,
always belong in pairs to the four interradial genitalia.

§ 143. Bursal genitalia of the Acraspedae (Tesseroniae). In all Tesseroniae the
central stomach remains free from the sexual productions, and the reproductive glands
are exclusively, or for the most part, developed in the subumbral wall of the four

Fig. P. Proc/iaragma prototypus (Cubomeduss, Charybdeidae).

Horizontal transverse section below the stomach, whose subumbral wall (gc) is completely visible, in
the middle, the oral opening («) with the four perradial oral lobes (al). The gastral filaments (/)
are placed upon the four interradial pyloric valves [gy). (gic) Subumbral wall of the two gastral
pouches (bp). (*) Genitalia, (ug) Gelatinous substance of the umbrella.

perradial pouches. In the most simple case, four interradial horseshoe-shaped glands
are formed, which include the four interradial cathammal nodes (hi) or the proximal ends
of the four narrow septal ridges (ks) in the concavity of their U-shaped proximal ends,
whilst the two limbs of each arch project into the two adjacent perradial pouches (PL
XV. figs. 2-6). The oldest and simplest of all Acraspedae, the Tesseridae (System, taf.
xxxi.), and some of the closely allied Lucernaridae (Halicythus), show this most simple
and apparently original condition. In all other Tesseroniae (and also in most Lucernaridae
and in all Peromedusae and Cubomedusae) each of the four interradial genitalia is divided
into two separate halves, as the convex proximal ends of the horseshoe-shaped gland (which
encloses the cathammal nodes) have undergone retrograde formation and disappeared,
so that only the two limbs remain. These lie on the two sides of the interradial

REPORT ON THE DEEP-SEA MEDUSA. ciii

cathammal septum belonging to them, but in two different perradial pouches, so that each
pouch contains the inverted halves of two adjacent pairs of genitalia. In the Luccrnaridse
(Pis. XVI.-XVII.) and in the Peromedusse (Pis. XVIII.-XXV.) the eight reproductive
glands, which are connected in pairs, lie as leaf-shaped swellings in the subumbral wall of
the four perradial pouches themselves. In the Cubomedusaa (PI. XXVI.), on the other
hand, they are only connected with the four perradial pouches by a narrow marginal
insertion (immediately next to the fused streak of the cathamma but on its subumbral
side), and otherwise project as eight free genital leaves into the hollow space of the
pouches, of which they occupy the larger part (fig. P, s ; System, taf. xxi.-xxvi.). In the
different Tesseroniae we find many stages of development in the structure of the
genitalia. In the simplest case, in the Tesseridse, the sexual glands are merely simple
ridges or pads, originated by wheel-shaped thickenings of the endoderm of the subumbral
wall (like the most simple vascular genitalia of the Craspedotse) ; a corresponding ridge
of the fulcral lamella serves as supporting frame (" sterigma ") for the subepithelial
germinal cells. Further (in part of the Peromedusse), more or less complicated folds are
formed, which rise above the subumbral wall and project freely into the space of the
pouches ; the supporting frame of the fulcral plate (" sterigma ") likewise rises higher
and attains greater development (Pis. XXIII. fig. 38, 39 ; Pis. XXV. fig. 5-7). In the
Cubomedusse the sterigma is developed into a broad, thin leaf, which is only connected
with the supporting plate of the subumbrella at the base of insertion (near the cathammal
ridge), and bears sexual cells on both its free surfaces (axial as well as abaxial) ; its freer
margin is sometimes lobed or branched dendritically (Chirodropus, System, taf. xxvi.).
The reproductive glands attain their most complicated structure in the Lucernaridse
(System, p. 386). Each of the eight reproductive leaves is here divided into numerous,
separate follicles, each containing a genital sinus with excretory passage, and each follicle
is sometimes again composed of a number of smaller sacs (PI. XVII. figs. 17-19).

§ 144. Gastral genitalia of the Acraspedse (Ephyronias). Whilst in all Tesseronia? the
subumbral wall of the four perradial pouches is the place of origin of the reproductive
glands, in the Ephyroniae or Discomedusse it has migrated centripetally from the pouch
wall to the subumbral wall of the central stomach. This centripetal change of locality must
therefore be phylogenetically considered as secondary, since the younger and more highly
developed Ephyronise are clearly derivable from the older and more simply constructed
Tesseroniae, and since the lowest and oldest grades of development of the former present
points of connection with the latter. In some Cannostonise (especially some Ephyridse)
in which the four primary interradial cathammal nodes are still preserved, we still find
four interradial horseshoe-shaped genitalia, which enclose the nodes by their concave
proximal arches and whose distal diverging limbs still lie in the subumbral wall of the
coronal sinus or of the four perradial pouches (System, pp. 467, 480, 492 ; taf. xxvii.,
xxviii., xxix.). The pairs of limbs are here often divided into separate halves, as the

CIV

THE VOYAGE OF H.M.S. CHALLENGER.

connecting proximal arch has become lost (Navphanta, PI. XXVII., XXVIII. ; Atolla,
PI. XXIX.). Whilst the first and oldest order of the Discomedusaj, the Cannostomse,
still present more or less the original genital conditions of the Tesseronise, these have
disappeared entirely in the other Discomedusae, in all Semostomse and Ehizostomse. As
here the four primary cathammal nodes are resorbed, and both the four perradial pouches,
separated by them and the coronal sinus are consequently merged in the flat central

Fig. Q. Zoncphyra 2>clagica (Discomedusa, Epliyrkla).

Subumbral view, giving the four orders of transverse axes (with twice as many radii). Tlio oral cross
(as) and the four perradial sense clubs (o') lie in the four perradii (Order I.). The gastral filaments
(/), the genitalia (s), and the four interradial sense clubs (o-) lie in the four interradii (Order II.).
The eight tentacles (to) and the tentacular coronal pouches (bt) lie in the eight adradii (Order III.).
The sixteen marginal lobes (?) lit' in the sixteen subradii (Order IV.). Sixteen bifurcate coronal
pouches, eight tentacular (bt) and eight rhopalar (bo) radiate from the central stomach (</).

stomach, four simple interradial genitalia usually lie in the subumbral wall of the central
stomach, lying nearer its centre in proportion as the Ephyronise character is more highly
developed and the peripheric umbrella corona becomes extended at the cost of the central
umbrella disk. The delicate, thin gastrogenital membrane (gg) on whose endodermal
inner surface the four genitalia are situated, is certainly simply termed the bottom of the
stomach or subumbral wall of the central stomach ; but it must always be borne in mind

EEPORT ON THE DEEP-SEA MEDUSAE. CV

that it is only its axial or proximal part which really deserves this name, whilst its
abaxial or distal part rather corresponds originally to the subumbral wall of the coronal
sinus and the four perradial pouches at its proximal margin, which have become merged
in the central stomach in consequence of the dissolution and resorption of the four
interradial cathamma. The delicate gastrogenital membrane is frequently (namely, in
the Pelagidse and Cyaneidaa) evaginated downwards out of the gastral cavity (Hke a
hernia), and forms four pendent gastrogenital pouches, in the bottom of which the four
frill-shaped genital bands lie (" extra versio gonadum," Pis. XXX., XXXI. ; System, p.
470, taf. xxx.). In most Rhizostomaa, on the contrary (and also in the Aurelida?), the
thin gastrovascular membrane invaginates inwards into the gastral cavity (like a replaced
hernia "intraversio gonadum," System, p. 470, taf. xxxiv.-xl.). If the four invaginated
inner gastrogenital pouches approach each other in the centre so that they touch, they
may become fused together and enter into communication by breach of the fused walls.
In this way there arises the remarkable formation, which has been already described, of
the " subgenital vestibule," of which the roof is formed by the cruciate gastrogenital
membrane, the floor by the brachial disk (comp. above § 96, p. lxxvii., woodcut, figs.
G and H, and PI. XXXII. ; also in the System, pp. 471-473, taf. xxxviii.-xl.). These and
other modifications of the gastrogenital formation of the Ephyronise have already been
described comparatively in the System der Medusen (1879, pp. 467-473). They are all
derivable from the four simple horseshoe-shaped genitalia, which we find in the subumbral
wall of the periphery of the stomach (or of the coronal sinus) in Ephyra, the oldest and
simplest form of the Discomedusae (fig. Q). In this and other morphological respects,
Ephyra remains one of the most important types among all Medusae.

(ZOOL. CHALL. EXP. — PART XII. — 1881.) M 0

EEPORT ON THE DEEP-SEA MEDUSAE. 1

DESCRIPTION OF THE SPECIES.

Class I. CRASPEDOT^E, Gegenbaur, 185G.

CRYPTOCARPiE, Eschscholz, 1829. GYMNOPnTHALMiE, Forbes, 1848.
Hydromedus*, Carus, 18G3. Aphacell.e, Hreckel, 1878.

Medusae without gastral filaments or phacellaa ; with ectodermal genitalia (or sexual
products formed from the external germinal layer) ; with a true velum (always without a
velarium) ; without true marginal lobes of the umbrella; with double centralised nerve-ring.
Phylogenetic descent (probably universal) and ontogenetic descent (now established for
the majority) derived from hydroid-polyps without gastral filaments or from hydrostomEe.
Ontogenesis chiefly alternation of generations, often with metamorphosis. The sexual
craspedote generation is formed by lateral gemmation from the asexual hydrostoma
generation.

Order I. ANTHOMEDUSiE, Heeckel, 1877.

Craspedotse without marginal vesicles and otolites, with ocelli at the bases of the
tentacles. Genitalia in the external or oral wall of the stomach. Number of the radial
canals almost always four, very rarely six or eight. Ontogenesis chiefly alternation of
generations, often with metamorphosis. The trophosome of the asexual generation is a
hydroid-polyp of the order Tubularise.

Family, Maegelid^ Hseckel, 1877.

MAEGELiDiE, Hasckel, System der Medusen, 1879, p. 68, taf. v., vi.

Anthomedusas with four or more simple or branched oral styles, with four or eight
separate genital sacs in the wall of the stomach, with four narrow simple radial canals, and
with simple unbranched tentacles, which are sometimes distributed equally, sometimes
grouped in four or eight bundles.

Sub-family, Thamnostomid^e, Hseckel, 1867.

Margelidee with branched or compound oral styles, and tentacles equally distributed,.
not grouped in bundles.

Thamnostylus,1 Hseckel, 1879.

Margelidse with branched or compound oral styles and only two opposite perradial
tentacles.

1 &xfivo;, bush ; tnv?.o:, style.
(ZOOL. CIIALL. EXP. — PART XII. 1881.) M 1

2 THE VOYAGE OF H.M.S. CHALLENGER.

The genus Thamnostylus with the single species Thamnostylus dinema (PI. I.) is the
only deep-sea Anthomedusa which I found among the collections of the Challenger
expedition. It belongs to the family of the Margelidse, to the sub-family of the Tham-
nostomidse, and is the only dissonematous genus of this sub-family, with only two
developed opposite marginal tentacles (like Cubogaster among the Cytaeidse).

The genus Thamnostylus is distinguished by the remarkable development of the oral
organs ; the long central oesophagus, which projects below far out of the central stomach,
and the four strong numerously branched oral styles which spring from its basis are
much larger in proportion to the rest of the body in Thamnostylus than in the other
Margelidse. In other respects Thamnostylus dinema appears at the first glance a very
abnormal and peculiarly formed Anthomedusa. Closer consideration and comparison with
other Craspedotse shows, however, that its structure is not special or peculiar to itself.
We rather find in it a peculiar combination of striking characters, which appear other-
wise combined in other Anthomedusa?. The Margelidse Limnorea triedra and Favonia
octonema described by Peron (1809) appear to come nearest to it (Peron, Tableau des
Meduses, No. 8, Annales du Museum d'Hist. Nat., torn. xiv. p. 329). Leseur has given
a very good figure of both (in pi. iii. of his Recueil des Planches inedites des Meduses,
figs. 3, 5), which clearly shows a near relation to Thamnostylus and Nemopsis. Here
the oesophagus also projects out of the umbrella cavity, and is surrounded by a bush of
blood red much-branched oral styles which spring from its basis. On the ground of
these figures (which have been copied by Blainville (1834) and Milne-Edwards (1849),
L. Agassiz has placed Limnorea and Favonia among the Rhizostomata. (Compare my
System der Medusen, 1879, p. 87, and also the figure of Nemopsis heteronema,
p. 93, taf. v. figs. 6-9).

Thamnostylus dinema, Hseckel (PL I.).

Thamnostylus dinema, Haeckel. 1879, System der Medusen, p. 85, No. 95.

Umbrella hemispherical, twice as broad as high. Stomach quadrangularly pyramidal,
reaching almost to the plane of the velum. Genitalia four egg-shaped swellings in the
wall of the stomach ; oesophagus quadrangularly prismatic, twice as long as the stomach,
projecting far out of the umbrella cavity ; four oral styles a little shorter than the
oesophagus, springing from its basis, 6 to 8 times dichotomised ; two opposite long,
strong tentacles, several times longer than the breadth of the umbrella. Horizontal
diameter of the umbrella, 16 mm. ; vertical diameter, 8 mm.

Habitat. — Antarctic Ocean, south from the Kerguelen Islands. Station 153. Lat.
65° 42' S., long. 79° 49' E. Depth, 120 fathoms.

The form of the umbrella is almost hemispherical, half as high as its greatest breadth
somewhat above the umbrella margin. The gelatinous substance thick, gradually and

REPORT ON THE DEEP-SEA MEDUSAE. 3

regularly thinning out towards the margin. The exurabrella (or the external convex surface
of the umbrella) has a finely punctured appearance, as small round, stinging papillae are
scattered equally over it (fig. 1). The subumbrella (or the inner concave surface of the
umbrella) shows a strong annular muscular system, and eight narrow radial or longi-
tudinal muscles, of which four perradial (mj)) accompany the four radial canals, and four
interradial run in the middle between the radial canals (mi). Their proximal ends pass
into the longitudinal muscles of the stomach and the (Esophagus. The velum projects
from the umbrella margin towards the interior as a somewhat broad muscular membrane,
and narrows the entrance to the umbrella cavity considerably. The umbrella cavity
itself is flat and limited in size as the .central third of it is occupied by the large gastral
pyramid with the reproductive sacs.

The umbrella margin is thickened into a roll, and pigmented red. At the points
where the four perradial canals open into the marginal circular canal, the rim of the
umbrella margin swells into four thick ocellar-bulbs with dark red pigment. Of these
the two opposite are without tentacles, whilst the two others, alternating with the former,
bear very long and strong tentacles (fig. 1). These are several times longer than the
diameter of the umbrella, cylindrical, thickened like a club at the base, and beset with
rings of thread-cells along the entire length.

Gastrovascular system. The central part consists of the quadrangularly pyramidal
central stomach, in whose wall the reproductive sacs lie; and of the projecting oesophagus,
double the length of the central stomach, from whose basis spring the four rnultibranched
bunches of the oral styles. The peripheric part of the gastrovascular system consists of
the four perradial canals which spring from the basis of the central stomach and open into
the circular canal at the umbrella margin ; a canal passes from the latter into each of the
two tentacles, and traverses its whole length. These two tentacle canals, as well as the
four radial canals and the circular canal uniting them, are rather narrow and ribbon-
shaped, and show nothing special. On the other hand, the central part of the alimentary
apparatus has a somewhat complicated construction.

The central stomach (figs. 1-3, gc) has the form of a quadrangular pyramid whose
height is nearly ecpial to the diagonal of its basis, and whose truncated end, which
has a downward direction, is the starting-point of the long oesophagus and of the four
tree-like oral styles. The quadrate basis of the quadrangular stomach pyramid occupies
the central third of the subumbrella, and is formed by the lower surface of the gelatinous
umbrella ; the four radial canals open into the central cavity at the four angles of the
square, and pass thence in the form of semi-cylindrical grooves to the four perradial
corners of the gastral pyramid. The thickened wall of these grooves forms the midrib of
the four leaf-shaped genitalia or pinnated "reproductive leaves."

Genitalia. Each reproductive leaf forms an egg-shaped swelling, with the rounded
basis turned upwards and the truncated point turned down. As four to five deep transverse

4 THE VOYAGE OF H.M.S CHALLENGER.

grooves on both sides of the midrib divide the genitalium into the same number of lobes,
it assumes the form of a delicate pinnated leaf. Its five to six pairs of pinnules become
longer from the top to the bottom and are delicately notched at the edges like many
fern-fronds (fig. 6). Numerous large and small ova appear lying very closely together
on the upper surface of the single pinnated leaves. The ova are large, naked amoeboid
cells, of irregular roundish or polyhedral outline, which enclose a large clear germinal
vesicle. In this nucleus a dark germinal spot of considerable size (nucleolus) is visible
which contains a distinct nucleolinus (fig. 8). The four reproductive leaves occupy
the greater part of the thickened wall of the central stomach, so that only four narrow
interradial area? of its external surface are free from them (fig. 3). More minute
investigation, however, shows that only the perradial midrib and the aboral basal
parts of the reproductive leaves are integrate portions of the wall of the stomach itself,
from the ectoderm of whose angles they are developed. On the other hand, the oral
points of the genitalia and the larger part of their lateral margins are free, and only
lie superficially on the external wall of the stomach. Between the colourless ova,
as in the remaining parts of the wall of the stomach and of the umbrella margin,
there are numerous fine granules of the same blood-red pigment (insoluble in spirit
of wine) which causes the red colour of the oral styles and of the stinging capsules of
the tentacles.

The oral styles (" stomostyli," figs. 1-5). In this species, as in several other Margelida?
{Hippocrene, Nemopsis, Bathkea, Limnorea, Sec), the characteristic oral styles form
extremely delicate multi-branched bunches, distinguished by their blood- red colour. These
branches are, however, more numerously and more strongly developed here in proportion to
the rest of the body than in all the other Margelidse. When fully extended (as it appears in
the particular specimen before us) they occupy a space exceeding that of the whole umbrella.
The four perradial strong stems of the oral styles are nearly as thick as the swollen basal
pieces of the tentacles and spring from the truncated point of the central stomach round
about the basis of the long oesophagus almost at the height of the plane of the velum
(fig. 1). Each of the four strong stems divides directly into two thick principal branches,
which again bifurcate after a short course. These branches appear to dichotomise at least
six or eight times (sometimes oftener), so that the aggregate number of the terminal
branches amounts to more than a thousand. The calibre of the branches becomes smaller
with each new bifurcation, so that the four basal principal stems are at least six to eight
times as thick as their terminal branchlets. Each of the latter ends with a spherical
stinging knob, which is composed of numerous longitudinally extended, radially placed
thread cells, and bears long fine cnidocils (fig. 5 «). The minute structure of the
oral styles and their branches is the same as in other Margelicke. Their principal mass
forms an endodermal cellular axis, consisting of a single row of flat coin-shaped endoderm
cells placed in series like a rouleau of money. Their nuclei, surrounded by an area of

KEPORT ON THE DEEE-SEA MEDUSAE. 5

protoplasm, lie in their centre (fig. 5 d), and distinctly form a connected axial cord in the
centre of the whole column of cells (which in the solid oral styles of other Margelidae, as
in the similar tentacles of the Narcomedusse and many hydroids, were erroneously con-
sidered by former observers to be a " central canal "). This solid endodermal cellular
axis is surrounded by a thin, but firm and very elastic, strongly refringent, supporting
lamella, by whose elasticity the extension of the contracted oral styles is effected on the
cessation of the contraction of the muscle (figs. 5-7, z). The muscles, which in a great
measure shorten and at the same time thicken the oral styles by their contraction, form a
thin lamella, composed of longitudinal, parallel fibres. This muscular plate, here as in the
tentacles, is a product of the ectoderm, whilst the elastic lamella is secreted by the endoderm.
The ectoderm covering of the oral styles consists of flat epithelial cells which partly form
flagellate capsules, partly stinging capsules, and also contain numerous granules of blood-
red pigment. The stinging cells of the end knobs are not pigmented.

The oesophagus or " gullet," which may also be termed " proboscis," and which springs
from the oral opening of the central gastral pyramid in the centre between the four stems
of the oral styles, is two or three times as long as the gastral pyramid, and projects far
beyond the velar opening of the umbrella cavity. It is quadrangularly prismatic, of equal
thickness in the two upper thirds, whilst the lowest third is swollen into an egg-shaped
pharynx without muscles. The latter is divided from the lowest part of the oesophagus,
which bears the quadratic oral opening, by a circular constriction ('c strictura palatina ")
(fig. 4). The thickened oral edge bears a circle of stinging knobs (figs. 4, an). The
four perradial corners of the oesophagus project strongly, whilst the concave lateral sur-
faces lie in folds.

Order II. LEPTOMEDUS^E, Hseckel, 1866.

Craspedota partly without, partly with marginal vesicles, these, when present, developed
from the insertion of the velum, with ectodermal otolite cells. Ocelli at the bases of the
tentacles sometimes present, sometimes wanting. Genitalia always in the course of the
radial canals. Number of the radial canals varying, sometimes four, six, or eight, sometimes
very large, sixteen, thirty-two, eighty, or even several hundred. Velum thin and delicate.
Ontogenesis, usually alternation of generations, often with metamorphosis. The
trophosome of the asexual generation is a hydroid-polyp of the order Campanulariae.

Family, Cannotid^e, Hseckel, 1877.

Caxnotid/E, Hseckel, System der Medusen, 1879, p. HO, taf. ix.

Leptomedusse without marginal vesicles, with four or six radial canals, which are
branched, forked, or pinnated, in whose course the genitalia lie.

6 THE VOYAGE OF H.M.S. CHALLENGED.

Sub-family, Polyorchid.e, A. Agass., 1862.

Cannotidae with four or six radial canals, which are pinnated, or furnished with
csecal side branches which do not reach into the circular canal.

Ptychogena,1 A. Agass., 1865.

Cannotidaa with four pinnated radial canals, whose alternating pinnated branches
bear several leaf-shaped cleft, indented or compound genitalia. Stomach a flat, wide
pouch, without special oral lobes.

The genus Ptychogena was established by A. Agassiz, in 1865, for the North
American deep-sea form, Ptychogena lactea (North American Acalephae, p. 137,
fig. 220). A second somewhat different deep-sea species from the North- Atlantic Ocean
{Ptychogena pinnulata) is here described, and completes Agassiz's short definition.
Ptychogena is the connecting fink between the apparently very different genera Gony-
nema and Staurophora. Whilst the stomach is a long tube in Gonynema, and is entirely
rudimentary in Staurophora, in Ptychogena it forms a flat, wide-opened quadrate pouch,
whose four corners pass conically contracted without definite limits into the four radial
canals ; and whilst in Gonynema the pinnated branches of the genitalia are entirely
limited to the radial canals, but in Staurophora run centripetally to the centrum of
the cross of these canals, they are developed in an intermediate degree in Ptychogena.
They there occupy only the proximal half of the radial canal, but pass from it some
distance upon the wall of the stomach. Both North Atlantic species of Ptycho-
gena appear to be true deep-sea Medusae. A. Agassiz writes of it as follows (loc. cit.,
p. 139):—

" This Medusa, like Tima, swims at a considerable depth below the surface. The action
of the light and increase of temperature of the surface is sufficient to kill them in the
course of half an hour ; the moment they are brought to the surface the spherosome loses
its transparency, the genital organs become dull, and the Medusa is soon completely de-
composed. This action is much more rapid than anything I have noticed even in Cteno-
phoras, Mertensia being the only genus in which the decomposing effects of light and heat
are at all equal to what is produced here. This Jellyfish must be a deep-water species, as
they have only been found during a single fall, and then only for a few days, when they
seemed quite abundant."

These remarks most probably are applicable to Ptychogena pinnulata as well as
Ptychogena lactea. The example of the former in the Challenger collection was found
at a depth of 1250 fathoms.

1 IIti/xi), turning ; y-vri, reproductive organs.

REPORT ON THE DEEP-SEA MEDUSAE. 7

Ptychogena pinnulata, Haeckel (PI. II.).

Pti/choijena pinnulata, Hreckel, 1879, System del Medusen, p. 148, Xo. 150.

Umbrella depressed, three to four times as broad as high. Stomach quadrangular, very
flat and wide, one-third of the diameter of the umbrella, with slightly-raised oral margin,
which is prolonged at the corners into four short lobes. Genitalia, four broad, almost
circular, pinnated leaves, which occupy the proximal half of the radial canals, at whose
conically-enlarged origin they pass into the wall of the stomach, each leaf with twenty to
thirty pairs of alternating pinnated branches, which are not divided, and which bear
leaf-shaped, deeply-notched reproductive lobes at the lower free margin. Two to three
hundred long tentacles with numerous marginal clubs between. Horizontal diameter of
the umbrella, 50-60 mm. ; vertical diameter, 20-30 mm.

Habitat. — North Atlantic Ocean. I was able to investigate several well-preserved
specimens in spirit of this North Atlantic species from the Copenhagen Zoological
Museum, which had been found by Captain Moberg between Ireland and Iceland (lat. 59°
7' N, long. 13° 32' W. from Greenwich). A fragment of a Cannotid, which 1 dis-
covered in a jar of the Challenger collection, from Station 50 (May 21, 1873), in the
same jar as Pectyllis arctica (dredged near Halifax from a depth of 1250 fathoms, lat.
42° 8' N, long. 63° 39' W.), appears identical with these specimens. Although this
decomposed fragment hardly included the quadrant of a disc, it was still sufficient to
identify it completely with these Copenhagen specimens from which the following descrip-
tion and drawing are taken.

Ptychogena pinnulata shows, on the whole, the same formation of the umbrella as
the closely-related Ptychogena lactea (loc. cit.). The umbrella is depressed, projecting
somewhat more strongly in the centre. The horizontal breadth at the opening of the um-
brella cavity is from two to three times as great as the vertical height. The gelatinous
substance of the umbrella is tolerably firm, but thin, and diminishes in thickness rather
rapidly from the centre towards the margin ; in the centre its thickness amounts to
5 or 6 mm. The exumbrella is smooth, without special distinct characters.

The umbrella margin is thickly beset with two rows of appendages, an outer row of
long tentacles, and an inner row of short marginal clubs (figs. 3, 4). The number of
tentacles or marginal filaments amounts to from 200 to 300 ; in one specimen I counted
320. There are usually from 70 to 80 upon each quadrant. They lie thickly pressed
together. The swollen basis or tentacle bulb is 1/8 mm. long, 0-6 mm. broad, and has
the form of a half-oval leaf lying in the meridian plane. The abaxial margin is strongly
arched, rising gradually from the basis, and then falling off rapidly. The axial margin is
straight or sinuated a little concavely. The marginal filament itself is very thin, and in
the spirit specimens before me nearly as long as the diameter of the umbrella ; in the
living animal probably three or four times as long. The marginal clubs, or tactile clubs

8 THE VOYAGE OF H.M.S. CHALLENGER.

(figs. 3, 4, oJc), lie on the inner side of the tentacles between their insertion and the
basis of the velum. In some places they alternate with the tentacles, but are usually
irregularly distributed. Their number appears very variable. In one of the three speci-
mens before me there are very few (20 to 30), in the second above 100, and in the third
over 200. The opaque marginal clubs appear chalk-white in reflected light, black in
transmitted light. They are pyriform, gradually enlarging from a narrow stalked base
about one-third to one-half as long as the basal bulbs of the marginal filaments, 0-6 to 0-8,
at the most 1 mm. long and 0"3 mm. broad. Treated with acid, they show a narrow
canal (fig. 8). The velum (r) is rather broad, but very thin and delicate, and with
many folds. The system of circular muscles of the subumbrella is moderately developed,
and shows no special peculiarities. The umbrella cavity is very shallow, its upper half
filled up in a great measure by the stomach and the four genitalia.

The wide oral opening leads into a short, shallow gastral sac, whose four basal
corners are extended into four conical funnels. These pass into the four radial canals,
which are pinnated in their proximal half and bear the genitalia (figs. 1, 2). The narrow
radial canals open at the umbrella margin into an annular canal, which sends branches
into the tentacles and marginal clubs (fig. 8).

The oral opening (figs. 1,2 in the centre) is quadrate, very wide, with irregularly
frilled borders, extending at the four perradial corners into four short wavy oral lobes
(fig. 2, al). The thin transparent wall of the quadrangularly prismatic gastral tube
hangs down nearly to the middle of the umbrella cavity ; its lower free oral margin is
much thinner. A perradial cross (fig. 2, g) whose four limbs are 0-5 mm. broad and
8 mm. long, appears very plainly in the fundus of the stomach on the gastral surface of
the gelatinous umbrella. This cross is formed by four very narrow ciliated grooves which
are centripetal processes of the umbrella wall of the four radial canals. In one of the
three examples the four limbs of the cross meet in the aboral centre-point of the sub-
umbrella, so that the quadrate ground of the stomach is divided into four congruent
equally limbed triangles (fig. 2). In the other two examples the points of the two
opposite triangles are truncated and rest (at one specimen at the length of 2 mm. in the
other at 6 mm.) in such a way that opposite points of the two other alternating
triangles remain at about the same distance from each other (fig. 7). The geometrical
form of the ciliated cross is here plainly amphitect, whilst it is completely regular in the
first specimen (fig. 2).

The four perradial corners of the fundus of the stomach are prolonged into four
conical funnels (fig. 2, ck) whose ends extend to the middle of the genitalia and
occupy the proximal half of the radial canals. The latter are very broad in the proxi-
mal half, and, on the other hand, very narrow in the distal half below the genitalia. In
the middle of its course each radial canal gives out a number of alternating pinnated
branches, twenty to thirty on each side, at right angles at the two edges (fig. 5). These

REPORT ON THE DEEP-SEA MEDUSAE. 9

branches are longest in the middle (up to 10 mm.), and gradually decrease in size towards
the ends of the radial canal; they are very short at both ends. The genitalia proceed
from their lower wall. The four genital glands (figs. 5, 6) seen from the subumbrellar
surface have the shape of broad elliptical or almost circular leaves, occupying the centre
of the radial canals from which they hang freely into the umbrella cavity (figs. 1,2). On
closer inspection it is shown that each gland forms a delicately-pinnated leaf, whose
20 to 30 pairs of pinnated branches are lamellae placed perpendicularly (s). The upper
edge of each lamella is connected with a transverse pinnated branch of the radial vessel,
whose lower (subumbral) wall forms a fold; the lower free edge of each lamella is
serrated or rather, split up into a number of finger-shaped points. The number of these
points is greatest (10 to 15) in the broadest lamellae in the middle of each gland, smallest
in the narrowest lamellae at the two ends of the genitalia. Each lamella is therefore, in
itself the half of a pinnated leaf, whose straight upper edge has grown to the transverse
branch of the radial canal, whilst the divided lower edge projects freely into the umbrella
cavity.

Order III. TKACHOMEDUStE, Haeckel, I860.

Craspedotae with auditory clubs, which sometimes stand freely on the umbrella margin,
and are sometimes enclosed in auditory vesicles, with endodermal otolite cells. Ocelli at
the basis of the tentacles usually wanting. Genitalia always in the course of the radial
canals. Number of the radial canals, sometimes four, sometimes six, sometimes eight,
never more ; between these there are often centripetal blind canals. The velum is firm
and broad. Ontogenesis, so far as we yet know, hypogenesis (or direct development
without alternation of generations) usually accompanied by metamorphosis.

Family, Teaciynemidj:, Gegenbaur, 1856.

Trachynemida;, Haeckel, System der Medusen, 1879, p. 255, taf. xvii.

Trachomedusae with eight radial canals, along whose course the eight genitalia lie ;
with long tube-like stomach without ventral peduncle ; with auditory clubs, which are
seldom free but usually enclosed in auditory vesicles lying on the umbrella margin.

Sub-family, Pectyllid.e, Haeckel, 1877.

Trachynernula with sucking cups on the tentacles, with radial mesogonia or broad
mesenterial bands.

(ZOOL. CHALL. EXP. PART XII. 1881.) M 2

10 THE VOYAGE OF 1 1. M.S. CHALLENGER.

Pectyllis,1 Haeckel, 1879.

Trachynemidse with eight genitalia in the course of the eight radial canals, without
centripetal canals. The genitalia are cut in two by eight radial mesogonia or leaf-shaped
mesenterial bands, and connected with the bases of the stomach. Oral cavity without
oral funnels, and without side pouches. Tentacles with sucking-cups very numerous
and placed closely one over the other in several rows on the umbrella margin. Numerous
(8 or 16 ?) auditory clubs.

The genus Pectyllis, with the two following genera Pedis and Pectantlris, form the
special small group of the Pectyllidse, which I placed in the System der Medusen
(1879, p. 265) as a sub-family in the family of the Trachynernidae. The Pectyllidse are,
however, so strongly distinguished by several striking peculiarities from the remaining
Trachynernidae, the Marmanemidse, that it is better to separate them as a special family.
The Pectyllidas agree with the Marmanemidse in having eight radial canals and eight
genitalia hanging from them, in the form of the depressed umbrella, and in the absence
of a gelatinous gastral peduncle ; on the other hand, they are distinguished from them by
two peculiar characteristics which are wanting in the other Trachornedusse, in having eight
mesogonia and numerous sucking-tentacles furnished with terminal sucking-cups. The
auditory clubs of the umbrella margin are free as in the Aglauridae, not enclosed in
"marginal vesicles" as in the Marmanemidse. The peculiar "mesogonia," or genital
mesenteries, are thin, membranous, vertically-placed leaves, which extend in the radial
plane between the central oesophagus on the one side and the eight sac-shaped repro-
ductive glands on the other, are inserted in the middle line of the latter, and sometimes
pass along the radial canals almost to the umbrella margin. The upper part of the umbrella
cavity is, therefore, divided into eight radial sections (" infundibula subumbralia ").
The peculiar sucking tentacles of the Pectyllidse are hollow or solid, very elastic and
contractile threads, which bear a powerful sucking-cup at the free end, and are used for
adhering by suction. Part of them resemble the " ambulacral-feet " of the echinoderms
in form and in the mode of motion. The sucking tentacles are very numerous in all
three genera of the Pectyllidse, sometimes closely packed together in several rows, one
above the other, on the margin of the umbrella ; sometimes grouped more or less distinctly
in separate bunches ; in all of them we can distinguish sixteen (or 32 to 64) bunches more
or less divided by marginal incisions, so that the umbrella margin appears almost lobed.
A further peculiarity of the Pectyllidse is the extremely broad, powerful velum, which
apparently, in all three genera, can be extended till, like a sphincter, it completely closes
the umbrella cavity ; they surpass all other Craspedotse in this extreme development
of the velum. The Pectyllidse are finally distinguished by a peculiar formation of the

1 ILjxTvAAi?, Pectyllis, derivative of Pedis.

REPORT ON THE DEEP-SEA MEDUSA. 11

circular canal, namely the glandular folds or tufts of its endoderm lining, which projecl
from its lower edge into the lumen of the vessel. As all the three genera of the Pectyllidae
which have been observed as yet (each only with one species) are typical deep-sea Medusse,
it is probable that the special peculiarities above mentioned are partially or entirely the
result of adaptation to life at great depths of the sea. At any rate they are of sufficient
importance to justify the separation of the Pectyllida) from the Marmaneroidse as a
special family. Among the three genera of the Pectyllidae, Pet-taut his is distinguished
by the sucking tentacles being grouped in sixteen separate bunches, whilst in the other
two genera they are closely crowded together along the edge of the umbrella margin.
Pedis is characterised by having blind centripetal canals which are wanting in Pectyllis
and Pectanthis.

Pectyllis aretica, Hasckel (Pis. III., IV.).

Pectyllis aretica, Hreckel, 1879, System der Medusen, p. 2G6, No. 287.

Umbrella nearly hemispherical, about 1^ times as broad as high. Exumbrella with
sixteen projecting radial ribs, alternating with sixteen deep radial grooves. Stomach
quadrangular prismatic, about as long as the radius of the umbrella. The margin of the
mouth fleshy and thickened, quadrate, with four perradial pointed lobes and four inter-
radial strong longitudinal muscles. Eight egg-shaped genitalia in the proximal half of
the radial canal, divided in two by broad radial mesogonia. Border of the umbrella
thickened, thickly beset with several rows of sucking-cups, which are divided into sixteen
larger and forty-eight smaller groups (16 to 20 sucking-cups in each group). Between these
arc placed sixteen longer and thirty-two shorter tentacles, and numerous (?) (8 to 16)
auditory clubs. Horizontal diameter of the umbrella, 18-24 mm. ; vertical diameter,
12-16 mm.

Habitat. — Arctic part of the North Atlantic Ocean.

I had an opportunity of examining several well-preserved spirit-specimens of this
species from the Copenhagen Zoological Museum, which were collected by Olrik in 1868
on the west coast of Greenland. A specimen from Station 50 of the Challenger list
(dredged 21st of May 1873, near Halifax, at a depth of 1250 fathoms; lat. 42° 8' N.,
long. 63= 39' W.) is identical with them. The Pectyllid described by Allman in 1878
as Ptychogastria polans (Nare's Narrative of a Voyage to the Polar Sea, vol. i. p. 299)
appears closely related to this species. I saw an incomplete specimen of this in 1879,
in the British Museum.

The umbrella (PL III. figs. 1, 2 ; PI. IV. figs. 3, 4) is nearly hemispherical, so that the
transverse diameter across the opening of the umbrella cavity is nearly twice as great as
the vertical axis (fig. 1). The former measures 18-24 mm., the latter 12-16 mm. The
gelatinous substance of the umbrella is thin but firm, and appears to be nearly of the

12 THE VOYAGE OF H.M.s. CHALLENGER,

same thickness throughout its whole extent (fig. 3). The exumbrella or the outer
convex surface of the umbrella is divided by sixteen rim-like radial ribs whose distal
ends project lobe-like on the margin of the umbrella, into sixteen deepened radial area? or
grooves (PL III. figs. 1,2); eight of the sixteen radial ribs correspond with the under-
lying genitalia and mesogonia (four perradial and four interradial). The eight principal
ribs unite above in the centre of the exumbrella into an eight-rayed star, while the eight
remaining adradial ribs which alternate with these do not reach the top of the umbrella.
Each two ribs are united at the distal end by a two-limbed clamp ; whose intercostal centre
turns inwards and downwards (figs. 1, 2). This arching inwards forms the boundary
In 'tween each two of the sixteen flat marginal lobes, and lies in the ideal prolongation of
the sixteen intercostal radial grooves of the exumbrella (" sulci exumbrales "). These form
the deepest part of the concave grooves between each two ribs. They are only sharply
defined in the middle zone of the exumbrella, and obliterated below and above. A very
delicate radial rib runs in the exumbrella between each groove and each rib (fig. 1).

The umbrella margin (" margo umbralis," figs. 1, 2, 4, 10, &c.) is considerably thickened,
and even at a superficial view appears divided, more or less clearly, into sixteen flat,
slightly convex, projecting lobes, each of whose central points forms the distal end of a
radial exumbral rib. The entire margin of the umbrella is continuously beset with nume-
rous tentacles (above a thousand), which are placed one over the other in several rows,
and are transformed into short-stalked sucking-cups. Isolated auditory clubs (8 or 16 ?)
are placed between them. A very broad velum projects inwards from the ring of cnida?
of the umbrella margin.

The number of marginal tentacles amounts to 1000 or 1200 ; all bear a terminal
sucking-cup. Only a small number of them appear in the form of longer or shorter
filaments ; in the greater number the filaments part of the tentacle is so much reduced,
and the terminal sucking-cup so strongly developed, that they appear as short-stalked
or even unstalked suckers (" acetabula"). The distribution of these appendages is very
regular and peculiar. Each of the sixteen marginal lobes bears a principal group, com-
posed of three larger sucking-cups and three triangular sucking-plates alternating
with them, and in each sucking-plate we can distinguish sixteen to twenty sucking-
cups of different sizes (comp. figs. 1, 2, 4, 10). Sixteen very large sucking-cup-
of the first size are the keystones of the arches, at the end of the sixteen exumbral
ribs (fig. 7, xa; fig. 10, A). Thirty-two sucking-cups of the second size lie some-
what depressed between them (fig. 8, xb ; fig. 10, B). Between these two series placed
still deeper and with longer stalks there are forty-eight suckiug-cups of the third
size (fig. 5, xc ; fig. 10, C). Finally, under these, closely crowded together in a
triangular group, are numerous smaller sucking-cups with longer stalks, their number
increasing towards the insertion of the velum, while their size diminishes (fig. 6, xd).
Each of these so-formed " sucking-plates" has a great resemblance to the sucking-plates at

REPORT ON THE DEEP-SEA MEDUSAE. 13

the end of the two long prehensile arms of the ten-armed Cephalopods (Sepia, Sec). It
was not so easy, unfortunately, to determine the number of the longer tentacles with a
small sucking-cup at the end, as most of these had been torn away. It appears, however,
that sixteen longer tentacles (hardly as long as the radius of the umbrella) are inserted
under the sixteen uppermost sucking-cups of the first size, and thirty-two shorter tentacles
between them below the sucking-cups of the second size (figs. 1, 2). The structure of
these tentacular formations is the same everywhere. The principal mass forms a strong,
solid axis of large chordal cells (or clear vesicle-like endoderm-cells, yt). This is
covered by a thin but firm and very elastic fulcral plate, and above this a thin layer of
longitudinal muscular fibres. The external epithelium of the ectoderm covering the
latter is rich in cnicke, which are usually accumulated on the abaxial side of the tentacle
basis in the form of a thick cushion of cniclse (figs. 6, n ; 8, n). The sucking-cup at the
truncated end of the tentacles is likewise furnished on its enlarged margin with a thick
ring of cnidae, and has a strong muscular plate on its concave surface (fig. 7, .<").
The mode in which these sucking-cups are very regularly and delicately distributed on
the margin of the umbrella will be best understood by comparison of figs. 1 and 2 in Plate
III., figs. 4 and 10 in Plate IV. and of the sections of the umbrella margin in figs. 5-8
(comp. the explanation of Pis. III. and IV.). The velum (figs. 2, 3,v ; 4,v), is very broad
and powerful, and appears capable of completely closing the umbrella cavity. The internal
axial half, whose free margin surrounds the narrow entrance to the umbrella cavity, is
much thinner than the strong external or abaxial half; the two halves are divided by a
deep annular fold projecting into the umbrella cavity (figs. 3, 4). The muscles of the
velum form numerous delicate circular folds. The subumbrella has also very strong-
circular muscles forming numerous ring-like folds on the whole lower surface of the
umbrella. The umbrella cavity is divided (as in Pectanthis asteroides) into eight deep
funnel-like sections, as eight broad, vertical septa (four perradial and four interradial)
stretch across from the eight radial canals and genitalia to the base of the stomach
(mesogonia, fig. 3, wa).

The central mouth leads into a tube-shaped quadrangular stomach, from whose basis
in the bottom of the umbrella eight radial canals pass out (four perradial and four inter-
radial). These bear the sac-shaped genitalia in their proximal half, which are fastened
to the subumbrella by the leaf-shaped mesogonia just mentioned. The stomach is a
very thick-walled tube with strong muscles, has the form of a quadrate pyramid, is fixed
by the narrow basis in the centre of the subumbrella, and hangs down in it nearly to
the middle of the height of the umbrella cavity (fig. 3, gp). At the quadrate oral open-
ing the four perradial corners of the pyramid (ah) pass into four triangular oral lobes
(<il) whose axial internal surface is divided by a (perradial) longitudinal groove into
two lips (fig. 2). An interradial thickening of the longitudinal muscle appears sharply
defined between each two oral lobes (figs. 2-11). The interradial oral edges which are

H THE VOYAGE OF H.M.S. CHALLENGER.

divided in two by the longitudinal muscles art; turned over outside (like the brim of
a hat). The very powerful muscles of the mouth and of the stomach he in numerous
folds projecting internally (figs. 2, 12).

The eight narrow radial canals (fig. 3, cr) which run at equal distances from the
fundus of the stomach (fig. 3, go) to the periphery, are there opened into a narrow
annular canal (fig. 3, cc ; figs. 5-8, cc). In the transverse section these walls show
numerous thick folds and tufts (<.!])), which proceed from the distal margin of the
canal, and project freely into its lumen. As usual, the outer or umbra! wall of the
annular canal, which lies close to the marginal portion of the gelatinous substance of
the umbrella (//), is covered by a flat, tabular epithelium (dn), whilst, on the other
hand, the remaining portion or the subumbral wall is covered by high, cylindrical
epithelium (dw). The latter also lines the tufts or folds of the lower wall of the
canal. At their basal end (which is turned from the lumen of the canal), the very
high cylindrical cells of the tufts contain granules of black pigment, enclosing a
nucleus. These pigmented tufts (which resemble the intestinal tufts of the verte-
brata) have probably secretive or excretive functions. (Comp. PI. IV. figs. 5-8, and
explanation.)

The eight genitalia or reproductive glands (figs. 2-3, sc) hang in the proximal halves
of the eight radial canals, as visible sacculations from it (four perradial and four inter-
radial). These appear as wide, fluted, thick-walled, egg-shaped, or fusiform sacs. A
radial mesogonium (or genital mesentery) rises in the middle of the subumbral wall of
each sac, as a thick vertical fold of the subumbrella, which passes from the basis of the
stomach to the umbrella margin (fig. 3, wr). These eight mesogonia attach the corners
of the stomach, halve the eight genitalia, and divide the space of the umbrella cavity into
the eight above-mentioned umbrella funnels, or the peripheric niches of the umbrella
cavity.

Pedis,1 Haeckel, 1879.

Trachynernidaa with eight genitalia in the course of the eight radial canals, between
which blind centripetal canals run from the annular canal. Genitalia connected with the
basis of the stomach by eight radial mesogonia or leaf-shaped mesenterial bands. Oral
cavity with eight invaginated oral funnels, and sixteen evaginated side pouches.
Tentacles with sucking-cups very numerous, closely crowded together in several rows
above each other on the umbrella margin. Numerous (8 or 16 ?) auditory clubs.

The genus Pedis is strikingly distinguished from both other genera of Peetyllidre by
its blind centripetal canals, which run out in large numbers from the annular canal (as
in Olindias among Petasidee, Glossoconus and GJossocodon among the Liriopida;, Carmaris

1 n.r,KTi:, hardened, stiffened.

REPORT ON THE DEEP-SEA MEDUSAE. 15

and. Cavmai ina among the Caxmarinidse). The sucking-tentacles beset the whole um-
brella margin, closely crowded in several rows over each other, as in Pectyllis, not in
separate bunches as in Pectanihis. The radial mesogonia are much less strongly de-
veloped than in the two other genera. The structure of the oesophagus, with its eight
curious adradial, ectodermal, oral funnel cavities, and the eight pair of endodermal side
pouches alternating with them, is quite peculiar, and, as far as I know, does not exist in
any other Craspedota.

Pedis antarctica, Hseckel (Pis. V., VI.).

Pedis antarctica, Hreckel, 1879, System tier Medusen, p. 2G6, No. 288.

Umbrella almost hemispherical ; about 1^ times as broad as high. Exumbrella
finely radially ribbed, with thirty-two more prominent ribs towards the margin.
Stomach quadrangularly prismatic, nearly as long as the radius of the umbrella. Oral
cavity with eight pairs of hemispherical side pouches. Oral margin fleshy and thickened,
quadrate, with four radial oral tentacles. Eight genitalia, egg-shaped sacs having
folds, in the proximal half of the radial canals, connected with the base of the stomach by
eight narrow radial mesogonia. 11 to 13 blind radial canals (3 larger and 8 to 10 smaller)
between each two radial canals. Umbrella margin, thickened into a roll ; thickly beset
with numerous rows of sucking-cups (about a thousand), forming thirty-two connected
groups. Between these numerous (8 to 16 ?) free auditory clubs. Horizontal diameter of
the umbrella, 36 mm. ; vertical diameter, 24 mm.

Habitat. — Antarctic portion of the Indian Ocean, S.S.E. from Kerguelen Island.
Lat. 60° 52' S., long. 80° 20' E. Depth, 1260 fathoms. Station 152. 11th February
1874. From this habitat I had only one single specimen for examination, but it was
complete and well preserved.

The umbrella showed, in a perfectly uninjured state, the peculiar natural shape
represented in figs. 1 and 2, Plate V.; fig. 11, Plate VI. A deep exumbral circular
furrow runs externally round the umbrella, about half-way up its height, dividing it into
an upper, nearly hemispherical, umbrella cone, and a lower, shallow, funnel-shaped,
umbrella margin. The largest transverse diameter of the umbrella (at the opening of the
umbrella-cavity) amounted to 36-40 mm., and was nearly double that of the largest
vertical diameter (in the axis of the umbrella).

The exumbrella (or the external, convex surface of the umbrella) is traversed
throughout by a very large number of fine radial ribs, amounting in the periphery of the
umbrella to 500 to 600 (PI. V. fig. 1 ; PI. VI. fig. 20). Besides these there are also
thirty-two more prominent radial ribs, running the whole length, and thirty-two less
prominent " costse exumbrales " (fig. 20, cs) alternating with them.

The gelatinous substance of the umbrella (fig. 2, ug) is nearly of equal thickness in
the upper aboral half of the umbrella, and nearly as thick on the radial section as the

16 THE VOYAGE OF H.M.S. CHALLENGER.

j-i

genitalia fixed on its subumbral side, It is very much thickened in the centre of the
apex, and projects into the fundus of the gastral cavity in the form of a short, conical,
gelatinous appendage (fig. 2, nk). On the other hand, the gelatinous substance is much
thinner in the lower oral half, hardly one-fourth or one-fifth as thick as in the upper
half, from which it is sharply divided (half-way up the height of the umbrella) by the
exumbral circular furrow. An enormous number of sinous elastic fibres, running from
the exumbral to the subumbral wall, traverse the gelatinous substance of the umbrella,
to which they give a considerable degree of firmness ; they are placed together in dainty
pyramids (fig. 8, iif), whose points touch the exumbrella (e), and their bases the sub-
umbrella (w) ; these pyramids form regular longitudinal series, corresponding to the
exumbral radial ribs.

The umbrella margin (" margo umbralis," figs. 1, 11, 12, 20) is not so visibly lobed in
this genus of the Pectyllidaj as in the two others, but rather appears to be of equal thick-
ness all over, and closely beset with an enormous number of short sucking-tentacles
placed in several rows above each other. Closer consideration, however, shows that this
garniture of the umbrella margin is by no means equally distributed, but rather arranged
in eight larger and thirty-two smaller groups. These, however, hang closely together,
and are not separated by depressions as in Pectyllis and PectardMs. A similar lobed
formation of the umbrella margin also exists fundamentally in Pedis, though it is not so
apparent externally as in the other two genera. Each of the thirty-two small groups of
tentacles (fig. 20, td) consists of from 30 to 40 solid tentacular appendages. Of these
the 16 to 20 upper (proximal) are short-stalked, pyriform, or club-shaped sucking-cups,
the 12 to 16 lower (distal) on the contrary are somewhat longer tentacles, partly with a
terminal sucking-cup, partly apparently forming feelers. These " tactile tentacles " with-
out sucking-cup (fig. 17) were mostly torn away, the longest barely above 1 mm. in
length. They are probably much longer in the living animal (as in Pectyllis arctica,
PI. III. fig. 1). The numerous sucking-cups form G to 8 alternating rows placed one
above the other on the umbrella margin ; in each of the thirty-two small groups, which
have an almost rhomboidal outline, they are placed in 5 to 6 diagonal rows, each with
4 to 5 tentacles (fig. 20). The size of the sucking-cups, which enclose a visible conical
ectodermal sucking cavity, decreases gradually from above downwards (fig. 12). A
larger sucking-cup is placed above somewhat further on the umbrella margin (fig. 20,
st). All the tentacles of this genus are solid ; their endodermal axis consists of large
clear chordal cells which are sometimes placed in a discoid row one behind the other
(fig. 17, dt), sometimes more numerous and in a more complete arrangement beside
each other (fig. 15, dt). This vesicular axial tissue is covered by very powerful internal
annular muscles, which thicken at the end into a strons; annular swelling ; single bundles
of external longitudinal muscles, very much thickened at the exumbral side of the
tentacles, extend out above these annular muscles ; they run out below the sucking-

REPORT ON THE DEEP-SEA MEDUSAE. 17

cups into eight to twelve radial cords, which converge towards the centre of the suck-
ing-cup (figs. 9, 10).

The auditory clubs (probably eight or sixteen) lie on the axial side of the umbrella
margin, under the insertion of the velum, inside the lowest row of tentacles. After most
careful search, I was only able to discover two or three of them, very small, and of the
same construction as in the Aglauridse. The thin endodermal axis of the auditory club
(fig. 16) consists of a few chordal cells ((/), of which the last is expanded like a vesicle,
and contains a large spheroidal otolith with concentric layers (ol). The ectodermal cells
of the epithelium of the club bear very long and fine auditory hairs (oh).

The velum (figs. 11-14) is thicker in Pedis than in any Craspedote hitherto known,
and is distinguished by a very unusual development of the muscular system. The breadth
of the velum is so great that when fully extended it can probably close the entire
umbrella cavity like a sphincter. The external abaxial half of the velum contiguous to
the umbrella margin, is nearly as thick as the tentacles, and three to six times as thick
as the internal axial half, from which it is divided by a deep circular furrow (fig. 11,
right half). If we draw the free projecting internal margin of the velum carefully
towards the inside, we can bring it so near the centre as to make it probable that the
umbrella cavity can be completely closed by the velum being drawn over it, as in the
foregoing species. The following layers (from above to below in the natural position of
the horizontal extended velum) can be distinguished in horizontal sections through the
velum : — (l) the ventral or subumbral epithelium of the velum (viv), containing dark-
brown pigment similar to that of the subumbral epithelium ; (2) a considerably thick
layer of clear vesicular connective tissue (x) ; (3) the muscular plate of the velum which
projects into this connective tissue in the form of numerous highly-developed circular
folds, each fold sending out numerous secondary folds or shoots into the clear plate of
connective substance (x), so that it appears delicately pinnated in the transverse section
(fig. 13, mv) ; (4) a thin but firm elastic supporting lamella, which sends out processes
into the muscular folds (zv); (5) the dorsal or exumbral epithelium of the velum (ve).
The epithelial cells of the exumbral epithelium of the velum are much smaller and
flatter than those of the subumbral.

The ectodermal epithelium of the lower surface of the umbrella or the subumbrella
consists of cells of dark brown pigment, from which the milk-white walls of the canals
are sharply thrown out. The underlying annular system of the subumbrella forms
numerous compact circular folds. In Pedis the umbrella cavity is simple, without
subumbral funnel cavities, as the eight radial " mesogonia " or " genital mesenteries "
so strongly developed in Pedyllis and Pectanthis are by no means so complete here.
They are merely indicated by eight narrow subumbral folds, running from the basis of
the eight genitalia to the basis of the stomach (fig. 2, wr).

The gastrovascular system in Pedis has, on the whole, the same plan as that of

(ZOOL. CHALL. EXP. PART XII. 1881.) M 3

18 THE VOYAGE OF H.M.S. CHALLENGER.

Pectyllis, but is easily distinguished by the blind centripetal canals running from the
circular canal, and also by the peculiar side pouches in the periphery of the mouth, and
the oral funnels alternating with these (PI. V. figs. 2-5; PI. VI. fig. 11). The central
oesophagus hangs down from the fundus of the umbrella cavity till past the middle of
it in the shape of a quadrangular tube beset above the oral opening with eight pairs of
side pouches. The eight radial canals running out from the basis of the stomach unite
at the umbrella margin into a circular canal, from which run numerous shoi't blind cen-
tripetal canals (PL VI. fig. 11).

The quadrate oral opening (figs. 3, 4, al ; fig. 11 in the middle) is surrounded by
a very muscular protuberant oral margin, whose surface has strong folds of the circular
muscle (fig. 3, me). These folds are divided into four interradial groups by four
perradial longitudinal muscles, ending in the four short heart-shaped oral lobes
(fig. 3, al). Above this muscular oral margin (which is probably very ductile, and
adapted for adhesion by suction) appears a very prominent circle of sixteen side pouches
(" bursse buccales," figs. 2-5, bb). These form hemispherical, or, more properly, semi-
oval evaginations of the gastral wall, and hang together in pairs in such a way that eight
pairs appear as oral bifurcated terminal shoots of the eight gastral grooves (gs) ; of
these longitudinal grooves of internal wall of the stomach, which proceed above in the
fundus of the gastral cavity (figs. 3, 4, cr). Each two pairs of side pouches are divided
by a peg-shaped oral funnel (infundibulum orale, figs. 2, 4, 5, 10). These conical
adradial oral funnels are quite peculiar invaginations of the gastral wall, in a certain
measure " internal side pouches " unknown to me in any other Medusa. Their conical
cavity, which is ccecal at the point, is lined by the ectoderm, and opens into the umbrella
cavity, whilst these " external side pouches" (bb) are lined by the endoderm, and open
into the oral cavity. Such " external side pouches" are unknown to me in any other
Craspedota, but they occur in the Periphyllidae among the Acraspeda (compare below).
The ectodermal external wall of the external side pouches is coloured with violet-brown
pigment, and has a broad milk-white dentated longitudinal striation in the middle.
The endodermal covering of the oral funnels is coloured milk-white, and sharply con-
trasted with the dark violet periphery of the oral cavity. The upper part of the gastral
cavity into which the gelatinous sphere of the umbrella (fig. 2, vb>%) projects, appears
eight-rayed in the transverse section (fig. 6), as eight adradial longitudinal folds running
from the eight oral cavities project inwardly between the eight concave gastral groups
into the lumen of the central cavity (g). Compare the perradial longitudinal section
(fig. 4).

The eight radial canals (figs. 11, 20, cr) which run from the basis of the stomach to
the umbrella margin, and there open into the circular canal, as well as the circular
canal itself (cr), and the blind centripetal canals proceeding from it, are not cylindrical
tubes, but flattened band-like vessels which are sharply distinguished by then milk-white

REPORT ON THE DEEP-SEA MEDUSAE. 19

colour from the dark violet subumbrella. In transverse section (fig. 12, ce) the circular
canal shows a high circular fold projecting nearly half-way into the lumen at its lower
marginal edge, which touches the basis of the velum (yc); the high cylindrical partly-
pigmented epithelium of this fold has probably a glandular nature like the similar folds
and tufts in the annular canal of Pectyllis.

Pedis is specially distinguished by the blind centripetal canals proceeding from the
annular canal, as these are wanting in the closely-related genera Pectyllis and Pec-
tanthis, although they are indicated in the latter by the eight adradial projections of the
circular canal, which alternate with the eight radial canals (PI. VII. figs. 11, 20, ce). On
the whole, the centripetal canals of Pedis comport themselves like those of Olindias among
the Petasidse, and of Glossoconus and Carmarina among the Geryonidas, but they are
shorter and broader, and have the form of a pointed equilateral triangle (fig. 20, ce).
Their aggregate number amounts to 80 to 100, as 11 to 13 centripetal canals lie between
each two radial canals, with the starting-point of their broad basis touching the
circular canal. Although their number and arrangement is not perfectly regular, the
(primary) adradial centripetal canal (in the middle between each two traversing
radial canals) is always the largest. Then follow the (secondary) centripetal canals,
which lie in the middle between the former and the latter, whilst the remaining- canals
are considerably smaller and irregularly distributed (comp. PL VI. figs. 1 1 and 20 ;
PI. V. fig. 2).

The eight genitalia (PL V. fig. 2, bs ; PL VI. fig. 11) in Pedis as in Pectyllis
(PL IV. fig. 3) are wide, folded, thick-walled pouches, which occupy the proximal half
of the radial canals and communicate with the lumen of the eight radial canals by a
wide fissure. The eight mesogonia or " genital mesenteries," which connect the genitalia
with the aboral half of the oesophagus, as broad radial lamellae in Pectyllis and
Pedanthis, are rudimentary in Pedis (fig. 2, ivr). In the only specimen examined (a
female) the wide cavity of the reproductive pouch was empty for the most part, and
only contained a few ova.

Pedanthis,1 Hasckel, 1879.

Trachynemidaa with eight genitalia in the course of the eight radial canals, without
centripetal canals. Genitalia halved by eight radial mesogonia or leaf-shaped mesenterial
bands, and connected with the base of the stomach. Oral cavity without oral funnels
and without side pouches. Tentacles with sucking-cups very numerous, divided into
sixteen separate bunches, each twro bunches between two radial canals; sixteen subradial
auditory clubs, one in the middle of each bunch of tentacles.

The genus Pedanthis is distinguished from the two other known Pectyllidae from

1 ri/i«T;j, firm, compact ; £i/ioc, a flower.

20 THE VOYAGE OF H.M.S. CHALLENGER.

the numerous tentacles being divided into sixteen isolated subradial bunches, so that
each bunch lies in the middle between a radial canal and an intercanalar adradial rib of
the exumbrella. A free auditory club is placed in the centre of each bunch upon a
projection on the lower side of the umbrella margin. The margin of the umbrella
appears distinctly divided by sixteen incisions (four perradial, four interradial, and
four adradial), into sixteen projecting lobes, each bearing a bunch of tentacles with
an auditory club. The eight mesogonia by which the umbrella cavity is divided into
eight funnels are strongly developed, as in Pectyllis, and the numerous caecal centripetal
canals between the radial canals, by which Pedis is distinguished, are also wanting as
in Pectyllis.

Pectanthis asteroides, Hseckel (Pis. VII., VIII.).

Pectanthis asteroides, Hseckel, System der Medusen, p. 267, No. 289.

Umbrella depressed to a hemisphere ; two to four times as broad as high. Ex-
umbrella with sixteen radial ribs. Stomach quadra ngularly prismatic ; nearly as long as
the radius of the umbrella. Mouth four-lobed, extensible into an octagonal sucking- disk.
Eight egg-shaped genitalia in the proximal half of the radial canal, encircling the basis
of the stomach in the form of an eight-rayed star, and halved by long radial mesogonia.
Umbrella margin swollen, thickened, with sixteen intercostal protuberances or marginal
lobes, each bearing a subradial auditory club and a pencil-shaped tuft of twelve to sixteen
tentacles. Tentacles hollow, of unequal length, the longest equal to the radius of the
umbrella, generally with a sucking-disk at the end. Horizontal diameter, 5 mm. ; ver-
tical diameter, 2 mm.

Habitat. — The Mediterranean. I myself caught a living specimen of this deep-sea
Trachomedusa with the tow-net in the Adriatic Sea on April 15, 1878, at a depth of
200 fathoms, some miles distant from Pola. The following description and the figures
on Plates VII. and VIII. are prepared from this specimen (a mature male), which was
examined alive. I also found a small specimen of the same species, which did not
admit of any minute investigation, but still furnished proof of its identity in a bot-
tom specimen from the Challenger collection from Station 4, at the entrance of the
Straits of Gibraltar. Lat. 36° 25' N., long. 8° 12' W. 16th January 1873. Depth, 600
fathoms.

In the example which I observed alive in Pola, the umbrella in a contracted condition
had an almost hemispherical bell shape. In a dilated condition, on the other hand, it
appeared distinctly depressed, so that the largest horizontal diameter was three to four
times greater than the largest vertical diameter. The former measured 4-5 mm., the
latter 1-2 mm.

The exumbrella or the external convex surface of the umbrella is divided by sixteen

REPORT ON THE DEEP-SEA MEDUSAE. 21

projecting radial-ribs into sixteen deepened radial areae or depressions, which project like
lobes on the umbrella margin (PI. VII. fig. 12). Of the sixteen radial ribs ("costse exum-
brales "), four perradial and four interradial (in the middle between the former) lie above the
mesogonia. These eight principal ribs are distinguished by thin streaks of puqile-red
pigment, which pass into eight large red ocellar spots at the umbrella margin, but are
wanting in the eight other alternating adradial ribs. All sixteen ribs are tipped with
nematocysts, which appear yellowish-white by reflected light and black by transmitted
light. These spots of pigment form a broad band in the periphery of the exumbrella
above the margin of the umbrella, and are divided from it by a colourless streak. The
eight red ocellar egg-shaped spots before mentioned are placed at the distal ends of the
eight red pigmented ribs ; they may, perhaps, be considered as true ocelli, though they
do not appear to contain a lens (comp. PI. VII. fig. 1). Besides these, there are also
sixteen large crescent-shaped golden-yellow spots at the marginal end of the ribs of
the exumbrella. The sixteen concave intercostal radial depressions of the exumbrella
alternating with these increase in breadth and depth towards the umbrella margin,
and are traversed in the middle by a deep-radial furrow (" sulcus exumbrahs "). A
pedunculated subradial auditory club lies at the end of these radial depressions (PI. VIII.
fig. 8, ok) in the middle of the projecting marginal lobe, and of the bunch of tentacles
borne by it.

The peculiar peripheric umbrella margin (" margo umbralis," PI. VIII. fig. 8) is con-
siderably thickened, and armed with a connected urticating ring {tic), consisting of a
thick accumulation of thread-cells. The sixteen protuberances or flat lobes are rounded
like an arch, and when looked at from above (and also at their subumbral surface) show
a cord of cilia pigmented black immediately inside the urticating ring. This ciliated
cord is sinous, and forms from eight to ten projecting vessels on each lobe (fig. 8, xp).
In the arching inwards of the umbrella margin between each two lobes, at the distal end
therefore of each exumbral rib, the black ciliated cord becomes a tongue-shaped pro-
jection, showing a funnel-like depression, which may perhaps be an organ of smell
(fig. 8, xo). A small free auditory club rises on the outer edge of each of the sixteen
marginal lobes enclosing a spheroid or elliptic otolite in the free end (in the last
endodermal cell, fig. 8, ok). The auditory club lies nearer the lower margin of the
umbrella, inwards from the insertion of the tentacles. More minute investigation was
unfortunately impossible.

The tentacles, which amount to 200 to 260, are divided into sixteen pencil-shaped
bunches, each two bunches between each two radial canals. In the transverse section the
tentacles are hollow (fig. 4), capable of great extension, movable and contractile, fur-
nished with a sucking-disk at the end, and are very similar to the ambulacra! feet of
the Echinodermata. The Medusae attach themselves by these sucking-disks to the
vertical walls of the glass vessel, and climb up them like an Asterias or a Sea-urchin

22 THE VOYAGE OF H.M.S. CHALLENGER.

(figs. 6, 10). The living specimen which I observed at Pola assumed moreover the
peculiar attitude represented in Plate VIII. fig. 7. The Medusa lies on its back, extends
a portion of the sucking-feet stiffly out round it, and attaches itself to the bottom of the
glass, whilst the other portion of the sucking-feet play freely in the water, as if feeling
and fishing for prey ; the mouth, therefore, stretches vertically from the opening of the
velum, which is contracted like a funnel, and also moves as if groping in different
directions. The tentacles, when extended, are almost as long as the radius of the
umbrella ; when contracted they are much shorter ; in the centre they are thickened like
a spindle, and become thinner at either end. A more minute investigation of the ten-
tacles shows that we can distinguish two different forms. The larger number have a
sucking-disk, which is pigmented red at the end, and are used for crawling and adhesion
by suction ; the smaller numbers are simply pointed at the end, without sucking-disk,
and are used as feelers, usually extended round and upwards, and moving like worms
(figs. 6-10).

The velum (fig. 3, vn ; fig. 7) is very thick, broad, and powerful. The plate of its
circular muscles is arranged in numerous circular folds projecting from the subumbral
surface, which, by interference, produce an iridescence. It is probable that in
Pectanthis (as in Pedis and Pectyllis) the velum can be extended so as to entirely
close the umbrella cavity like a sphincter. The circular muscles of the subumbrella form
projecting circular folds similar to those of the velum, but are arranged in sixteen arcades
corresponding to the sixteen exumbral ribs and to the sections of the umbrella margin
between every second lobe (fig. 9, wn). The subumbral exoderm is distinguished by
scattered nematocysts (fig. 9, wn). The umbrella cavity is divided, as in Pectyllis
arctica, into eight separate compartments, or funnel cavities, as the eight broad leaf-
shaped mesogonia (fig. 9, ivr) are stretched between the radial canals and the basis
of the stomach (see below).

The central four-lobed oral opening leads into a tube-shaped four-sided stomach, from
whose basis eight radial canals (four perradial and four interradial) run out in the
bottom of the umbrella cavity. These bear the eight genitalia as sack-shaped evagi-
nations in their proximal half, and are united in the umbrella margin by a circular
canal which sends out branches into the tentacles. The central oesophagus (fig. 9, gb)
hangs down in the middle of the umbrella cavity as a muscular tube of a gold-yellow
colour, and prismatic quadrate form. The four interradial lateral surfaces are slightly
depressed into a groove between the four rounded perradial bodies. The oesophagus
is nearly as long as the radius of the umbrella when extended, much shorter when con-
tracted. The oral opening is divided by four shallow grooves into four short blunt oral
lobes (al) armed with an accumulation of nemocysts. The mouth can project out of
the umbrella cavity through the opening of the velum, and extend externally in the
form of a very thin flat octagonal sucking-disk (fig. 3, am).

REPORT ON THE DEEP-SEA MEDUSAE. 23

In a more extended condition the extremely thin oral disk extends almost to
the margin, and shows eight small triangular lappets (four perradial and four inter-
radial, fig. 3, am) at the margin of the mouth. Eight narrow radial canals run from
the periphery of the bottom of the stomach ; these are united at the umbrella margin
into a circular canal, and the eight genitalia placed in their proximal third.

The eight genitalia (four perradial and four interradial) are egg-shaped, thick-walled
sacs, pigmented red, and with the side walls touching each other ; they surround the
basis of the stomach like an eight-rayed star (fig. 2, s ; fig. 9, s). The sacs are half as
long as the oesophagus, occupying the upper half of the umbrella cavity, and contain a
large evagination of the radial canal (fig. 5, sc). A radial genital mesentery or meso-
gonium, a vertical radial fold with brown stains of the subumbrella, is inserted in the
middle of the subumbral wall of each genitalium ; this begins at the basis of the stomach
and runs along the subumbral median bine of the radial canals to the margin of the
umbrella (fig. 9, wr). These eight mesogonial leaves fasten the edges of the stomach,
halve the eight genitalia, and divide the space of the umbrella cavity into the eight
peripheric niches (or imperfect funnel cavities) above mentioned. The transverse section
of each genitalium shows that they actually consist of two completely separated halves,
between which the basis of insertion of the mesogonial fold (tor) is intersected as a
dividing septum. The living specimen examined by me in Pola was a male. The two
sperm-sacs of each genitalium (sm) were divided by a strong fulcral plate (2) from the
high cylindrical epithelium of the cndoderm (d), and lay immediately under the endoder-
mal epithelium, to which they owe their origin.

Order IV. NARCOMEDUS^, Hasckel, 1877.

Craspedotae with auditory clubs, which always stand freely on the umbrella margin,
with endodermal otolite cells. Ocelli at the basis of the tentacles usually wanting.
Tentacles inserted dorsally, connected with the distant umbrella margin by peronia
which divide it into a number of collar lobes. Genitalia originally in the lower or
oral wall of the stomach, from which they often extend peripherically into the radial
gastral pouches. Radial canals sometimes wanting, sometimes present, in which case
they extend in the form of flat, radial, gastral pouches. Annular canal sometimes
obliterated, but always divided by the radial peronia into a number of arched canals
bordering the margin of the collar lobes. Number of the radial parts (tentacles,
lobes, and pouches) indefinite and vaiying, rarely four, usually eight or more up to
thirty-two. Ontogenesis, as far as we know, usually hypogenesis, often accompanied
by metamorphosis.

24 THE VOYAGE OF H.M.S. CHALLENGER.

Family Cunanthid^e, Hseckel, 1877.

Cunanthim:, Hseckel, System der Medusen, 1879, p. 310, taf. xix. figs. 1-3, taf. xx. figs. 1-6.

Narcomedusse with broad pouch-shaped radial canals or pernemal gastral pouches, which
are sometimes simple, sometimes split each into two cascal lobe pouches, but are always
connected with the circular canal by double peronial canals, with otoporpse (or clasps of
the cordyli) at the basis of the auditory clubs.

Sub-family, Cunoctonid^:, Hseckel, 1877.
Cunanthidse, whose radial pouches bifurcate each into two csecal lobe pouches.

Cunarcha? Hseckel, 1879.

Cunanthidse, with four perradial tentacles, inserted in the bifurcation of four perradial
gastral pouches, which at the distal part are continued into four lobe pouches.

The genus Cunarcha is one of the simplest and oldest of all forms of the Narcome-
dusse, and is immediately connected with Cunantha, the hypothetic originating genus of
this order ; like the latter it has only four tentacles and four alternating collar lobes,
but is distinguished from it by the four perradial gastral pouches being continued at
the distal end into two csecal lobe pouches ; eight csecal lobe pouches, therefore, lie in
pairs between the four tentacles in the periphery of the umbrella collar. This genus
forms thereby a very interesting phylogenetic transition between Cunantha and
AUgina. By retrograde formation of the otoporpse and the proximal part of the radial
pouches it would be transformed into ^Egina. The presence of lobe pouches is common
to Cunarcha, and to the genera Cunoctona and Cunissa ; they compose the special
sub-family of the Cunoctonidee, whilst the other sub-family of the Cunanthidse, the
Cunoctanthidse, have simple radial pouches without lobe pouches (Cunantha, Cunoc-
tantha, Cunina), System, p. 314.

Cunarcha aginoides, Hseckel (PL IX.).

Cunarcha ceginoides, Hseckel, 1879, System der Medusen, p. 315, No. 329.

Umbrella cap-shaped, one and a half times as broad as high. Lens of the umbrella
biconvex. Umbrella collar with four broad, oval lobes, as long as the radius of the lens.
Mouth with long conical oesophagus ; four gastral pouches very broad and short, deeply

1 Cunarcha. Oldest form of Cunoctonidoo.

REPORT ON THE DEEP-SEA MEDUSAE. 25

cleft into eight distal lobe pouches occupying the greater part of the collar lobes ; four
perradial tentacles longer than the diameter of the umbrella ; twelve auditory clubs
(three on each lobe), the middle auditory club twice as large as the two lateral.
Horizontal diameter, 4 mm. ; vertical diameter, 2 mm.

Habitat. — West Coast of Africa. I first observed a living specimen of these species
off Lanzerote, one of the Canary Islands, in December 1866, and the figures of Plate IX.
and the following description are taken from this specimen. Later I found a specimen
in a glycerine preparation of the Challenger expedition, containing the beautiful Phaeo-
daria Ccelodendru7ii, south of the Azores, west from the Canary Islands, lat. 32° 41' N.,
long. 30° 6' W. Depth, 1675 fathoms. This glycerine specimen from the Challenger
collection was very imperfectly preserved, but sufficiently preserved to settle its identity
with the living specimen found at the Canaries. It is, however, possible (or probable)
that this Medusa does not belong to such a great depth, but was captured in shallower
water in drawing up the lead.

The umbrella (PL IX. figs. 1-3) has the form of a flat cap, and is nearly once and a
half as broad as high. When the broad velum hangs loose, the aggregate height of the
umbrella (including the velum) nearly ecpials the greatest breadth (in the middle of
the height). The umbrella is divided into two distinct anatomical portions, the upper
" umbrella lens " and the lower " umbrella collar," by a deep horizontal circular furrow of
the exumbrella, the coronal furrow (fig. 3, ec). The central umbrella lens or umbrella
disk (fig. 3, iv) is simply formed by the gelatinous body of the umbrella, and has the
form of a thick biconvex lens with a rounded edge ; its upper surface is covered with the
flat exodermal epithelium of the exumbrella, and is somewhat more strongly vaulted
than the lower gastral surface, which is covered by the endodermal epithelium of the
stomach. The two surfaces are connected by numerous fine, sinuous elastic fibres
which traverse the gelatinous substance perpendicularly (fig. 6, uf; fig. 7, uf). The
consistence of the gelatinous substance is considerable, resembling that of a soft gelatinous
cartilage. The umbrella collar (" umbrella edge " or shortly " collar "), as we shall term
the portion of the umbrella lying underneath the insertion of the tentacles, has a very
complicate structure as contrasted with the simple lens lying above it. It consists of a
corona of four lobes arising from deep radial indentations or incisions of the umbrella
margin, the peronial furrows. And these, again, are caused by the four tentacles having
left their original position on the umbrella margin and having emigrated a little way
into the exumbrella. I consider this peculiar centripetal change of position of the
tentacles into the exumbrella, which is probably connected with their partial development
into feelers, as the first " true cause " of the manifold and varying transformations,
which the umbrella margin and the adjacent organs undergo in all Narcomedusse.
Originally the tentacles were placed immediately on the margin of the umbrella as in the
other Craspedotse. When they passed upwards into the external surface of the umbrella,

(ZOOL. CHALL. EXP. PART XII. 1881.) M 4

26 THE VOYAGE OF H.M.S. CHALLENGER.

they took with them an urticating streak of the urticating ring of the margin, and this
urticating streak becomes the umbrella clasp or " mantel clasp " ("peronium") (figs. 3,
4, en). This contains the basis of the tentacles in continuous connection with the
urticating ring of the umbrella margin, and so gives rise to a radial furrow, more or less
deep, of the umbrella collar which cuts through its gelatinous substance, but leaves the
subumbrella intact. Transverse sections show that the peronia are tolerably thick, solid,
clasp-like cords, consisting of a peculiar modification of urticating tissue ; numerous strata of
large spheroidal nematocysts lie closely accumulated one above the other. The distal end
of the umbrella clasp is in continuous connection with the urticating ring of the umbrella
margin (fig. 4, nc), whilst its proximal end is inserted at the spot where the tentacle
(t) runs out from its conical root (tr). The exumbral surface of each peronium lies
freely in the depth of the peronial furrow (between each two collar lobes), whilst its sub-
umbral surface is covered by the longitudinal muscle (fig. 5, mp) which rises from the
margin of the umbrella to the liases of the tentacle. Both lateral margins of the
peronium touch the adjacent peronial canals (ck).

The four collar lobes (" lobi collares," fig. 4), which are separated by the four peronia,
are of a broad uniform shape ; their broad proximal basis falls in the coronal furrow of
the exumbrella, whilst their distal point forms the usually projecting part of the umbrella
margin (figs. 2, 3). Its distal external margin, which bears three auditory clubs on each
lobe, is joined to the velum; the lateral margins are limited by the peronia. The lobe
pouches (with the sexual products) lie in the concave internal or axial surface of the
collar lobes, whilst the external or abaxial half is vaulted convexly. The thin gelatinous
plate of the lobes is considerably thickened in the middle, and a sharp edge of the
exumbrella projects like a rib in the interradial middle line (figs. 1, 4, er).

As in all Narcomedusae, the peculiar umbrella margin in Cunarcha ceginoides is
strongly indented in consequence of the dorsal change of position of the tentacles, and
supported by a strong urticating ring (fig. 4, nc ; fig. 7, nc). This consists, like the
peronia, of thickly accumulated nematocysts, and is covered with a ciliated sensitive
epithelium. The proximal margin of the urticating ring touches the annular canal (cc),
its distal margin rarely touches the velum (v). The two thin nerve rings (fig. 7, re' dorsal
or exumbral, re" ventral or subumbral nerve ring) which are separated from one another
by the supporting lamellae of the velum, lie on the two edges of the urticating ring.
The radial section (fig. 7) shows the more special anatomical and histological conditions
of the umbrella margin. (Comp. the explanation.) The umbrella margin of Cunarcha
(Bginoides has sixteen tentacular organs, namely, four long strong perradial tentacles,
four interradial large auditory clubs, and eight adradial smaller auditory clubs.

The four perradial tentacles, which are common to Cunarcha and the nearly related
Canantha, remain permanent only in these two genera, whilst in all remaining Cunan-
thidae this original number either increases or only amounts to four in the first early

REPORT ON THE DEEP-SEA MEDUSiE. 27

stage. As in all other Narcomedusse, the tentacles are solid and fastened in the
gelatinous substance of the umbrella by a peculiar " tentacle root." At the point of
insertion of the tentacle where the " root " runs into the gelatinous substance, both
tentacle and root are in continuous connection with the proximal end of the peronium,
whose distal end passes into the urticating ring of the umbrella margin. The muscle
and the nerve of the clasp, which maintain direct communication between the nerve ring
of the umbrella margin and the tentacles, run on the axial side of the peronium. We
may therefore say that the solid dorsally inserted tentacles are composed of three
essential parts, which join at the point of insertion, viz., (l) the tentacle filament or the
free projecting part ; (2) the tentacle root, which is enclosed as a support in the
gelatinous substance ; and (3) the peronium which maintains the connection with the
umbrella margin. The tentacle filament, or the free projecting part of the tentacle
(figs. 4, 6 1), shows precisely the same structure which we have already described in the
solid tentacles of the Pectyllidse. The endodermal axis, which originates from the endo-
derm of the circular canal, forms a cylindrical column and consists of a single row of
large, clear, discoid chorda! cells, lying one above the other like the coins in a rouleau of
sovereigns. The conical or carrot-shaped tentacle root (figs. 4, 6, Ir), a direct process
of the endodermal axis, projecting more or less into the gelatinous substance of the
umbrella, consists of similar cells. The point of it has a centripetal direction and lies
with its lower (umbra!) side on the upper (exumbral) side of the gastral pouch, which it
likewise serves to support firmly. A structureless septum divides it from the gelatinous
substance covering it, and from the adjacent endoderm of the vascular system. The
exodermal epithelium of the free tentacle filament, which consists partly of thread
cells, partly of sense cells, does not run from its insertion at the root, but passes con-
tinuously into the urticating epithelium of the peronium. The urticating cells, which
contain nematocysts, are tolerably ecpially distributed ; so are the sense cells, which
partly bear cilia or feeling bristles. At the club-shaped swollen distal end of the tentacles,
the spheroidal thread cells are more thickly accumulated, and the cilia of the sensitive
epithelium considerably prolonged so as to form a thick bunch (fig. 3). The part of the
insertion of the tentacle, where filament, root and peronium join, is surrounded as with a
collar by a thick semi-circular urticating swelling (figs. 2, 4, 6, n).

The twelve auditory clubs of this species, as in all Narcomedusae, must be regarded as
" modified acoustic tentacles " (System der Medusen, p. 307). The four interradial
(primary) auditory clubs Avhich lie on the point of the four corona! lobes, are from two to
three times as large as the eight adradial (secondary) (fig. 4, oV). The free projecting
lithocyst is club-shaped, and sits with a thinner short stalk upon a flat roundish " audit ory
pad " (figs. 4, 6, 7, 8). The solid axis of each auditory club consists of three to four short,
broad, discoid endodermal chordal cells, of which the proximal is the smallest, and con-
tinuously connected with the endodermal epithelium of the annular canal. The distal end

28 THE VOYAGE OF H.M.S. CHALLENGER.

cell of the axis, on the other hand, is very large, and encloses a prismatic crystalline
otolite (fig. 8, ol). The sense-epithelium forming the exodermal covering of the audi-
tory clubs is composed of very long, fine auditory cilia, which diverge radially and so
form a bundle in whose axis the club is placed (figs. 7, 8, oh). An auditory clasp, also
termed a " marginal mantel clasp " or " centripetal urticating streak " (" otoporpa," fig.
4, oo; fig. 8, oo), runs from the basis of each auditory club. It is a broad thickened streak
of the exodermal epithelium, consisting of a thick accumulation of thread cells and covered
with ciliated sense cells. The longitudinal axis of the auditory clasp is a centripetal pro-
longation (sometimes straight, sometimes broken) of the longitudinal axis of the auditory
clubs ; both lie in the same meridian plane. The auditory clasps of Cunarcha ceginoides
are of the same nature as those of Cunoctantha polygonia (System, p. 317, taf. xix.
fig. 2). They are shorter and stumpier than in most other Cunathidae, almost triangularly
club-shaped, gradually broadened from the thin distal end (at the marginal urticating
ring) towards the broad proximal end, and ending there in a thick crescent-shaped
urticating swelling similar to that at the insertion of the tentacle (figs. 4, 8, 0})). Like
the three auditory clubs of each collar lobe, the auditory clasps belonging to them are very
unequal in size, the medial (interradial) two to three times as long and broad as the two
lateral (adradial) ; whilst the latter only project slightly above the proximal margin of
the annular canal, the former extends till between the two lobe pouches of each lobe
(figs. 2, 3, 4, op).

In Cunarcha ceginoides, as in all other Narcomedusaa, the peculiar " subumbrella "
is limited to the concave ventral side of the peripheric umbrella collar, whilst the
entire ventral surface of the central umbrella lens is occupied by the broad gastral
disk. The circular coronal muscle of the subumbrella consequently forms a broad mus-
cular ring, which only lines the concave surface of the four collar lobes ; its upper or
proximal line of limitation touches the coronal furrow and the periphery of the stomach,
whilst its lower or distal margin is divided from the strong velum by the urticating
ring and nerve ring of the actual umbrella margin.

The velum is of considerable breadth, thick and compact, considerably broader at
the four perradial peronial indentations of the umbrella margin than at the four inter-
radial points of the collar lobes (fig. 1, 3, v); it is sometimes extended tensely horizontally,
and in that case it narrows the entrance to the umbrella cavity so much that only a
narrow opening for the passage of the oesophagus remains ; sometimes it projects down-
wards like a funnel (fig. 3, v), and sometimes it hangs loosely and vertically from the
umbrella margin like a compact multifold curtain. Like the four-lobed umbrella collar,
the broad velum as well as its distal process, present a very different appearance
according as they are dilated or contracted, and this is also the same with the umbrella
uavity, of which they form the wall (comp. figs. 1, 3, 6). The latter usually appears as
a narrow annular hollow space, whose internal (axial) wall is formed by the conical basal

REPORT ON THE DEEP-SEA MEDUSAE. 29

half of the oesophagus, and its external (abaxial) wall formed above by the umbrella collar,
below by the velum. The upper (proximal) margin touches the peripheric margin of the
gastral cavity, and corresponds to the exumbral coronal furrow, whilst the lower (distal)
margin forms the opening of the umbrella cavity.

The gastrovascular system (figs. 1, 4) is divided in Cunantha CBginoides, as in all
other Narcornedusad, into two essential and very distinct principal parts — into the
central stomach with oesophagus, and the peripheric corona of pouches with a circular
canal ; the former is fastened to the ventral side of the central umbrella lens, the latter
to that of the peripheric umbrella collar. The central stomach (gc) is a flat circular
pouch, whose horizontal covering or upper aboral wall is formed by the lower, slightly
convex, depressed surface of the central gelatinous lens of the umbrella. The bottom
or lower wall of the central stomach, on the contrary, is only a narrow ring, whose thick
muscular wall extends downwards like a cone, and becomes a long strong oesophagus
(fig. 3, gt). This oesophagus is very mobile and contractile, nearly as long as the
horizontal diameter of the umbrella, the upper half conically funnel-shaped, the lower
half nearly quadrangular])* prismatic ; it ends below in a narrow oral opening, which
sometimes seems quadratic, sometimes circular (fig. 3, na). Like the whole lower wall
of the stomach, the proboscis-like oesophagus is capable of great extension and con-
traction.

The peripheric corona of pouches which runs from the periphery of the basis of the
stomach (towards the inside of the coronal furrow) begins with four broad perradial
gastral pouches, lying crosswise (figs. 2, 4, bg), whose upper (adumbral) wall is sup-
ported in its perradial middle line by the stiff tentacle root lying on it (tr). The
breadth of the four gastral radial pouches increases remarkably towards the outside, and
surpasses the length considerably ; after a short course they bifurcate into two semi-
oval c-ecal pointed lobe pouches (figs. 2, 4, hi). These fill the largest part of the sub-
umbral wall of the collar lobes, and at the same time represent the genitalia, as the ova
are developed from the exodermal epithelium of their subumbral wall (fig. 4, so). Only
a few (two to four) large ripe ova lie in each lobe pouch, among numerous others very small
and undeveloped. A double clasp canal or peronial canal (fig. 4, ck) runs out between
the two lobe pouches of each perradial gastral pouch from the middle of the distal end
of the latter. This double canal consists of two narrow parallel tubes, which are separated
by the deep furrow of the peronium or umbrella clasp. The two parallel canals diverge
on the umbrella margin, at the distal end of the peronium, turn almost rectangularly in
contrary directions, and run along the distal margin of the umbrella lobe, in whose centre
they unite with the half of the corresponding neighbouring canal running towards them.
In this way there is formed a peculiar annular canal shaped like a garland or a festoon,
whose arches border the periphery of the umbrella lobes, and whose inverted corner
corresponds with the insertion of the tentacle. In proportion, as the arches of the lobe

30 THE VOYAGE OF H.M.S. CHALLENGER.

collar in the Cunanthidse becomes rounded, the shorter is the double canal, which
connects the free-arched margin of the annular canal with the radial gastral pouch, and
the longer at the same time the festoon canal, as we may suitably term the annular
canal, with its bow-shaped archings inwards. However isolated this formation of the
vascular system in the Cunanthidse may appear at first sight, it may easily be re-
ferred back to that of some of the closely allied Trachomedusse (especially the Geryonidse).
In these Geryorddae, in which the flat " genital layers," or pouch-like broadened radial
canals only reach to the annular canal, we only require to insert the indentation of the
umbrella margin deeper into the opening place of the annular canal, and to represent
the umbrella clasps as running straight through to the basis of the tentacles, in order
to have the formation of the Cunanthidse. The annular canal of the Geryonidse (and of
the remaining Craspedotse) consequently corresponds to the entire " festoon canal " of the
Cunanthidse, which is composed of the double peronial canals and the peripheric " mar-
ginal canal " (lying on the marginal urticating ring), which connects them. On the
other hand, the four broad perradial " gastral pouches " of Cunantlia and Cunarcha cor-
respond to the four typical " radial canals " of the Craspedotae, which are also sometimes
extended like a ribbon (as, for example, Liriopc and Glossocodon). The two lateral
wings of the last, in which the sexual products develop, may also be compared to the
two sexual "lobe pouches" of Cunarcha (comp. my System der Medusen, 1879, pp.
304, 306, taf. xix. fig. 2 ; taf. xx. figs. 1, 2). The paired lobe pouches of Cunarcha and
Cunoctantha (I.e., taf. xx. figs. 1, 2) are at the same time nothing else than the " inter-
nemal gastral pouches " of the iEginicke (I.e., taf. xx. fig. 11). Cunarcha therefore appears
to be a very interesting phylogenetic intermediate form between Cunantha and AZgina,
immediately connecting these two tetranemal parent genera of the Cunanthidae and
iEginidse with each other. (Comp. the tabular " Uebersicht liber die Homologien der
Kadial- Canal e einiger tetranemalen Trachomedusse and Narcomedusse " in my System der
Medusen, 1879, p. 336.)

Family, Peganthid^e, Hseckel, 1877.
PEGAKTHiDiE, Hteckel, System der Medusen, 1879, p. 323; taf. xix. figs. 4-7; taf. xx. figs. 14, 15.

Narcomedusse without radial canals and without gastral pouches in the subumbrella,
but with a festoon canal (or a circular canal formed by a circle of separate lobe canals),
with otoporpce or auditory clasps at the basis of the auditory clubs.

Sub-famdy, Polyxenid.e, Hseckel, 1877.

Peganthidae with a single circular genitalium, forming a simple or lobed girdle in
the subumbral wall of the stomach.

EEPORT ON THE DEEP-SEA MEDUSAE. 31

PolycoJpa,1 Hseckel, 1879.

Peganthidse with a simple reproductive girdle forming a broad ring in the lower wall
of the stomach (with genital cseca in the lobe cavities). Numerous (10 to 30) collar
lobes, and the same number of tentacles alternating with them.

PolycoJpa is the simplest, and phylogcnetically the oldest genus in the family Pegan-
thidse, that peculiar group of Narcomedusse which are distinguished by the complete
want of the radial canals, and by the formation of a bow-shaped festoon canal. "Whilst
the most closely related Cunanthicke have still several true radial canals (in the form of
broad pernemal "gastral pouches "), these disappear completely in the Peganthidse, inas-
much as the strong tentacles by their dorsal change of position occupy the whole of
the umbrella margin up to the periphery of the stomach, where they are inserted. The
original radial canal (the " gastral pouch ") as well as the peronial double canal, undergo
retrograde formation ; both become lost. The annular canal, however, is divided into
the same number of isolated vessels as there are lobes of the umbrella collar. Each
horseshoe-shaped lobe canal or bow canal edges the margin of its lobe, and opens at
base of the latter immediately into the gastral cavity (beside the insertion of the
tentacles) by two separate mouths. The whole gastrovascular system in the Peganthidse
therefore consists only of the flat lens-shaped gastral sac and the circle of isolated lobe
canals, each opening with two mouths into the periphery of the stomach. Of the four
genera of the Peganthidse, which are all closely related, Polycolpa appears to be the
simplest and oldest form, as it has the primitive formation of the genitalia. The simple
genital girdle forms a broad undivided ring in the lower or subumbral wall of the
stomach. It does not send out csecal or pouch-like processes into the separate lobe
cavities as in the closely related genus Polyxenia and in the genera Pegasia and
Pegantha, derived from the latter.

Polycolpa forskalii, Haackel (PI. X.).

Polycolpa forskalii, Hfeckel, 1879, System der Medusen, p. 328, No. 350.

Umbrella flat and discoid ; two to three times as broad as high. Twenty-five lobes,
nearly pentagonal ; none as long as broad. Genital girdle very broad, occupying nearly
the whole lower wall of the stomach. Twenty-five tentacles, three times as long as the
radius of the umbrella ; 130 to 170 auditory clubs (5 to 7 on each lobe). Horizontal
diameter, 20 to 30 mm. ; vertical diameter, 8 to 10 mm.

Habitat. — The Indian and Pacific Oceans. I myself observed a living (female)
specimen of this species in the Ped Sea. It was taken in the tow-net at a depth of
above 60 fathoms, and the figures in Plate X. are drawn from it. I recognised, as I

1 rio?ii/>!o>i7r«, with many curves (at the umbrella margin).

32 THE VOYAGE OF H.M.S. CHALLENGER.

believe, the same species in an incomplete fragment found by the Challenger expedition
near Mindanao, one of the Philippine Islands, at a depth of 82 fathoms. Station 201.
Lat. 7° 3', long. 121° 48' E. 26th October 1874.

I name this species in honour of the meritorious Swedish naturalist, Peter Forskal,
who not only gave the most trustworthy description of Medusa? in the last century, and
was the first to describe the Medusa? of the Red Sea, but also (in 1775) made the
first (and hitherto best !) description and drawing of a Peganthid (Pohjxenia mollicina).

The umbrella (PI. X. figs. 1-3) is depressed, discoid, nearly two to three times as broad
as high, and divided, as in all Peganthidas, by a deep horizontal coronal furrow (fig. 3, ec),
into an upper half, the massive umbrella lens, and a lower half, the lobed umbrella
collar. The thick umbrella lens ("umbrella disk" or "gelatinous mantel") consists of
a planoconvex or biconvex gelatinous mass of a cartilaginous or even caoutchouc-
like consistency. The solidity of the gelatinous disk, connected with a high amount
of elasticity, attains its maximum among the Craspedota? in this family. The cause of
this extreme solidity are the innumerable branched, net-like, anastomosed, elastic fibres
which run crosswise through the gelatinous substance from the external to the internal
surface of the umbrella. The vertical thickness of the umbrella lens is one-third as
great as its greatest horizontal diameter. The exumbrella is flat, without any special
distinguishing character (fig. 2). The umbrella collar, which is sharply divided from the
umbrella lens lying above it by the deep circular constriction, consists of a circle of
twenty-five thick gelatinous lobes, and of the broad velum, which not only completely
fills the interspaces between the lobes or the pernemal incurvatures of the subumbrella,
and connects them like a swimming membrane, but also projects inwards a considerable
way about the external margin of the lobes. The limits of the umbrella collar and the
umbrella lens is marked by a circular line, in which the tentacles are inserted, and in
which the openings of the festoon canal in the periphery of the stomach lie. (Comp.
figs. 2, 3, 6.)

The umbrella lobes — or more accurately " the gelatinous lobes of the umbrella collar "
— consist of a process of the gelatinous substance of the lens, which becomes thinner
towards the exterior in the direction of the margin of the lobes. Although the thick-
ness of the gelatinous substance in the lobes is not nearly so great as that of the central
lens, it is still considerable, and the lobes have great solidity. It is therefore difficult
to flatten out the marginal lobes, which are strongly rolled inwards both in the living
and the dead animal. The circle of rolled-up lobes makes the umbrella here (and still
more in other Peganthida?) look like the flower of the turncap lily {Lilium martagon).
The outline of the collar lobes is sometimes more rectangular, sometimes more pentagonal,
according to the state of contraction (figs. 1, 2, 6). The lateral margins, as well as the
point, is always strongly curved inwards ; its exumbral external surface is, therefore,
strongly curved both in a radial (longitudinal) and a tangential (transverse) direction.

RErORT ON THE DEEP-SEA MEDUS/E. 33

The subumbral internal surface is strongly concave, corresponding to this external
vaulting, and forms a protecting cavity, which is only open in a radial direction
towards the umbrella cavity. We shall designate these cavities, which are essentially
niches or secondary cavities of the umbrella cavity, and surround it like the altar niches
of a circular temple (Pantheon), the lobe cavities ; in most other Peganthidse (as in the
following species Pegantha pantheon) they serve for the reception and protection of the
genital sacs, which branch out from the gastral genital ring. The central umbrella cavity
itself (fig. 3, h), which is very flat and low as in all Peganthidse, is limited above by the
subumbral gastral wall and the genitalia lying in it, whilst it opens wide below (fig. 1).

The subumbrella is represented in the central part as far as the lower surface of the
umbrella lens by the muscular subumbral gastral wall, as this extends to the borders of
the lens and collar. In the peripheric part, on the other hand, at the lower surface of
the umbrella collar, the subumbrella forms a circle of isolated muscular plates lining the
inner concave surface of the lobes. The muscular ring of the subumbrella appears lobed
in the lower part, from the proper margin of the umbrella (with the nerve ring and
indicating ring), being deeply indented between every second lobe. The velum completely
occupies the interspaces between the lobes to the umbrella margin like a swimming mem-
brane, and moreover projects internally a little further than the connecting annular margin
towards the axis of the umbrella cavity. The velum is very thick and compact, laid in
many folds, and, like the lobes, almost always found more or less rolled up. Concentric
annular folds predominate in the inner or axial part of the velum, whilst radial folds
predominate in the outer or abaxial part, which runs in between the folds in the form
of a triangular tip (figs. 6, 8, v). The subumbrella is so deeply indented between
each two lobes that the triangular tips of the velum rise between them as far as
their base and the insertion of the tentacles (figs. 2, 6). The structure of the sub-
umbrella and of the velum is the same as in the next following species (comp. PI. XII.
fig. 12).

As the proper umbrella margin (in a morphological sense) is not determined by the
free axial margin of the velum (the limit between exumbrella and subumbrella), but
rather by the marginal urticating ring and the double nerve ring lying on it, the true
umbrella margin appears deeply indented in Polycolpa forshalii, as in all Peganthidse.
It forms a continuous margin of the collar lobes and, at the same time, the frontier line
between these and the velum (fig. 6, nc). The festoon canal lies on the inner margin
of the urticating ring (fig. 6, cf) which accompanies it all along. The more minute
structure of the umbrella margin is the same as in the following species (comp. PI. XII.
fig. 12). Of tentacular organs the umbrella margin bears twenty-five tentacles and a
large number of auditory clubs (five to seven on each lobe).

The tentacles, whose number in all Peganthidse ecpials that of the collar lobes, alter-
nate regularly with the latter, and are inserted at the bases of every two lobes in the

(ZOOL. CI1ALL. EXP. PAET XII. 18S1.) M 5

34 THE VOYAGE OF H.M.S. CHALLENGER.

coronal furrow of the exumbrella (figs. 1-3, 6). The twenty-five tentacles of our Pohj-
colpa forskalii are l| to 2 times as long as the diameter of the umbrella, and are some-
times curved upwards like a crown (as in Pegantha pantheon, PI. XL fig. 1), sometimes
turned downwards under the umbrella (PL X. fig. 2) ; they are cylindrical, cartilage-
like filaments, somewhat thickened at the bases, becoming gradually pointed finely
towards the end, and combining a high amount of stiffness and firmness with considerable
elasticity. The solid chordal axis resembles the chorda dorsalis of the vertebrata, and
consists of large, clear, thick-walled endoderm cells, which have a firm elastic membrane,
transparent contents, and a large nucleus. The chordal axis of each filament forms a
single row or column of such coin-shaped chordal cells (comp. PL XII. figs. 10, 11).
Its exodermal epithelium contains numerous spheroid nematocysts, especially in the
abaxial side of the filament. The clear conical tentacle root is also composed of thicker
chordal cells, it penetrates radially (centripetally) some way from the insertion of the
tentacles into the gelatinous substance of the disk, and often lies, bent like a hook, with
its lower oral side on the periphery of the stomach (PL X. fig. 3, tr ; fig. 7). A net of
branched protoplasmic filaments radiating from the layer surrounding the nucleus is
visible in each chordal cell of the root (fig. 7). The endodermal supporting plate,
which encloses the chordal axis of the tentacles like a tube, also surrounds the root up to
the point, which, on the other hand, the layer of longitudinal muscular filaments (lying
outside the endodermal supporting plate) does not. " Umbrella clasps " or peronia,
which appear so strongly developed in Cunarcha and JEginura, are only rudimentary
in Polycolpa and Pegantha as in most Peganthidse. As the deep indentations of the
umbrella margin extend between each two lobes almost to the base of the tentacle, the
peronia are naturally so much shortened that they almost disappear. By their retrograde
formation the insertion of the tentacle remains in continuous direct connection with the
urticating ring as it passes immediately into the former at the base of the lobes. The
interlobar points of the velum, therefore, also extend to the tentacle root between
each two lobes (fig. 6).

Pobjcolpa forskalii has 130 to 170 auditory clubs, 5 to 7 on each of the twenty-five
lobes (fig. 6). One of them is placed on the point of the lobe, the others (in pairs
opposite each other) on its lower lateral margin. Their structure is the same as those
previously described in Cunarcha. Here, however, each of the 3 to 4 endodermal
axial cells usually contains an otolite (fig. 8, ol). The otoporpse or " auditory clasps " at
their bases (fig. 8, oo) are club-shaped urticating streaks of the exumbrella covered with
ciliated sense-epithelium with larger and smaller nematocysts (fig. 8, n). Their other
functions are the same as in Pegantha pantheon (comp. PL XL fig. 4).

The gastrovascular system (figs. 1, 3, 6, 8) has the special formation, peculiar to all
Peganthidse, which distinguishes this family of the Medusae from all the rest. It con-
sists of two principal sections, the central stomach and the peripheric festoon canal (fig.

REPORT ON THE DEEP-SEA MEDUSAE. 35

6, cf; fig. 8, cf). The latter runs like a garland along the margin of the lobes and opens
throughout between every two tentacles with a double mouth in the periphery of the
stomach. The stomach is a completely flat, circular, or polygonal pouch, occupying the
entire lower surface of the umbrella lens (fig. 3, gc). Corresponding to the latter, the
upper wall or cover of the stomach forms a flat or only slightly convex, rarely concave,
circular surface, whose periphery presents in certain conditions of contraction a regular
polygon ; each of its projecting corners corresponds to a tentacle insertion, each side
of the base to the base of a collar lobe. The projecting corners sometimes form
triangular pouches with the ends directed towards the insertion of the tentacle (last
rudiments of radial pouches). The lower wall or bottom of the gastral pouch is a
circular or regularly polygonal thick, muscular plate, covered with endoderm above and
exoderm below. The oral opening, which is extended into a short cylindrical oesophagus
hanging freely down, is in the centre (fig. 3, at). The thickened oral margin is simple, not
split up into oral lobes. The muscular plate appear considerably swollen at the oral
margin (longitudinal section, fig. 5, m). Numerous gland cells (gd) are scattered
between the high cylinder cells of the gastral endoderm (dg), they are 2 to 3 times
as broad as the latter, have twice as large a nucleus, and are distinguished by the turbid,
granular nature of the protoplasm. As in all Narcomedusae, the muscular wall of the
stomach is capable of considerable contraction and dilatation.

The pecuhar festoon canal (" canalis festivals," fig. 6, cf; fig. 8, cf) which attains its
highest development in the Peganthidae and the complete want of radial canals con-
nected with it, suffices alone to characterise this family and to distinguish it from all
other Medusae. Phylogenetically this peculiar condition is simply derived from that of
the Cunanthidaj, and from the fact that the stomach stretches by peripheric growth as
far as the insertion of the tentacles (or to the limit of the umbrella lens and the umbrella
collar), and so includes the broad pouch-shaped radial canals. The deep sinuses which
are found in the Cunanthidas between each two radial pouches are in some measure
obliterated in the Peganthidae. Hence the " triangular points " of the periphery of the
stomach, which in some Peganthidae run out to the insertion of the tentacles (already
described by Eschscholtz in Polyxenia as " long three-sided processes of the stomach "),
must, in fact, be considered the last rudimentary remains of radial canals. While in the
Cunanthidaa the latter still serve to connect the stomach with the radial canal, in
Peganthidae the triangular points open into the periphery of the stomach in as many
places as there are insertions of the tentacles between each two collar lobes. The circular
canal has, therefore, the same disposition as in the nearly related Cunanthidae ; it
runs along the velar margin of the collar lobes immediately under the urticating ring
of the true umbrella margin ; it is, however, interrupted at the basis of each two ad-
jacent lobes by the insertion of the tentacle, and opens into the stomach beside the
latter. The state of the case may be expressed thus : the annular canal of the Pegan-

3(> THE VOYAGE OF H.M.S. CHALLENGER.

thidse is divided into many (10 to 20) bow-like or semicircular lobe canals, running on
the margin of the gelatinous lobes inside from the insertion of the velum, and opening
at the bases of the lobes into the periphery of the stomach. The circular canal is, how-
ever, in no way reduced but forms a spacious tube, whose lumen in the larger species
is often a millimeter in diameter and allows the introduction of a probe. Its
end'oderm is usually thickened and laid in folds, in some species even rising into
numerous tufts or papilke (like intestinal tufts). No such folds are recognisable in the
transverse section of the festoon canal in our Polycolpa forskalii, whilst in the species
immediately following a low annular fold is clearly present at the distal margin of the
canal as in Pedis (comp. PL VI. fig. 12, yc, and PI. XL fig. 12, yc).

The reproductive glands in Polycolpa forskalii appear in their simplest form as a
broad circular girdle, occupying the largest part of the lower subumbral gastral wall
(fig. 1, sf; fig. 3, sf). In the living female specimen observed by me in the Red
Sea, this girdle was of a beautiful sky-blue colour, so were the tentacles and the urticat-
ing ring of the umbrella margin ; the points of the tentacles were dark-blue. The
subunibral convex external surface of the reproductive girdle is tolerably smooth, and
only traversed by insignificant and incomplete radial folds (fig. 1, sf). A thick com-
pressed mass of small egg-cells, between which isolated large ova are scattered, appears
in the radial transverse section (fig. 3, sf; fig. 4), between the high gastral endodermal
epithelium of the ova (fig. 4, dg) and the flat subumbral exodermal epithelium (fig. 4,

Sub-family, Pbgasid^e, Hasckel.

Peganthidfe, with a circle of several separate genitalia, forming dilatations of the
subumbral gastral wall and lying apart in the lobe cavities of the umbrella collar.

Pegantha,1 Hseckel, 1879.

Peganthidae, whose gastral reproductive girdle is divided into a circle of separate
vesicle-shaped genital sacs, equal in number to the tentacles and alternating with them
(a simple or multi-lobed caecum in each lobe cavity). Numerous (10 to 30) collar lobes,
and the same number of alternating tentacles.

The genus Pegantha represents the most complete and phylogenetically the youngest
genus of the remarkable famdy of the Peganthidaa, in which the famdy type reaches its
highest development. Whilst in the preceding genus Polycol'pa, the ancestral genus of
the family, the genitalium appears as a simple girdle in the lower wall of the stomach,
which, in Poh/ccnia and Pegasia develops into a peripheric circle of lobes, in

1 Hnyti, a spring ; autln, a flower.

REPORT ON THE DEEP-SEA MEDUSAE. 37

Pegantha the girdle is divided into a circle of completely separate perigastral genital
sacs, one of each hanging freely in every lobe cavity of the umbrella collar (comp.
System der Medusen, 1879, p. 327, pi. xix. figs. 4-7).

Pegantha 'pantheon, Hseckel (Pis. XL, XII.).

Pegantha pantheon, Hreckel, 1879, System der Medusen, p. 332, No. 359.

Umbrella crown-shaped, twice as broad as high, eighteen egg-shaped lobes, one-half
as long as broad. In each lobe cavity a simple genitalium in the form of a broad
roundish delicately-twisted leaf, eighteen tentacles twice as long as the radius of the
umbrella, 400 to 450 auditory clubs (23 to 25 at each lobe). Horizontal diameter, 20
mm. ; vertical diameter, 10 mm.

Habitat. — The South Pacific Ocean, near Mindanao, Philippine Islands. I found the
extremely well preserved (male) specimen of this species, from which the figures in
Plates XI. and XII. are taken, in the same bottle of the Challenger collection which con-
tained the fragment of the preceding species. Station 201. Lat. 7° 3' N., long. 121° 48'
E. 26th October 1874. Depth, 82 fathoms.

The umbrella (PI. XL fig. 1 ; PI. XII. figs. 7-9) is shaped like a diadem or crown ;
is nearly twice as broad (20 mm.) as high (10 mm.), and divided by a deep horizontal
coronal furrow into a massive upper half, the umbrella lens, and a lobed lower half, the
umbrella collar. The massive upper part or umbrella lens consists, as in the pre-
vious species, of a tolerably firm biconvex gelatinous lens, of which the horizontal
diameter is twice as great as the thickness (fig. 7, tig).

The exumbrella is distinguished by branched strongly-projecting ribs, between which
deep radial furrows traverse the external upper surface (figs. 1-8). The ribs of the
umbrella lens, which increase in thickness from the centre towards the periphery, are
distributed so that a thicker principal rib runs in the middle of each collar lobe from
which several thinner secondary ribs branch out laterally.

The umbrella collar consists of a circle of eighteen oval umbrella lobes (figs. 7, 8).
These "gelatinous lobes of the umbrella collar " were closely pressed together in the
specimen examined, and were so strongly rolled inwards, and of such a cartilaginous con-
sistency that they could only be opened out flat under strong pressure (fig. 8, right half).
They then appear of a broad oval, once and a half as long as broad. The convex
external surface of each lobe (fig. 2) is strongly vaulted and traversed by five projecting
longitudinal ribs of which the centre rib is considerably thicker than the lateral. The
concave inner surface of each lobe (fig. 3) encloses a roomy lobe cavity, in which hangs
a genital sac with folds. These eighteen lobe cavities (fig. 3, hi) form a circle of niches
or secondary cavities round the central umbrella cavity, and surround it like the altar-
niches of a circular temple (Pantheon). The central umbrella cavity itself is flat and

38 THE VOYAGE OF H.M.S. CHALLENGER.

low, opening wide below, whilst it is limited above by the subumbral gastral wall (conip.
figs. 1, 7, 9).

The suburnbrella, with its annular muscular layer is divided, as in the previous
species, into two very different parts, limited by the circle of genitalia (fig. 9). The
central part of the suburnbrella, which corresponds in extent to the lower surface of
the umbrella lens, is formed by the lower, folded, very muscular gastral wall, and appears
pierced in the middle by the oesophagus (fig. 7). The peripheric part, on the other
hand, is composed of the circle of isolated muscular plates which line the inner concave
surface of the eighteen collar lobes. The velum completely fills the narrow interspaces
of these lobes, and, moreover, projects freely a little way further over the points of the
lobes like a connected circular edge (fig. 1, ve, right half v). Vertical sections of the
firm velum (fig. 12, left) show that the upper (subumbral or ventral) epithelium of the
velum (vw) is three times as high and as thick as the lower (exumbral or dorsal) epithe-
lium (mo). A strong circular muscular layer lies on the former (mv), and a thick elastic
supporting plate on the latter (zv).

As in the other Peganthidse, the peculiar umbrella margin (characterised by the urticat-
ing ring and nerve ring) is deeply indented, and covers the selvage of the collar lobes
like a connected edge. In Pegantha pantheon it forms eighteen deep curves reaching
as far as the insertion of the tentacles (figs. 2, 3, 8, 12). In the radial transverse
section of the umbrella margin (PI. XII. fig. 12) the urticating ring (na) appears
covered by dense epithelium with long cilia. The dorsal nerve ring (re) is divided
from the ventral nerve ring (re") by the supporting lamella of the velum (zv) ; both lie
immediately outside (abaxially) the insertion of the velum (comp. the explanation of
fig. 12). The distal margin of the broad festoon canal (cf) touches the velum imme-
diately. Of tentacular organs, the umbrella margin bears eighteen tentacles and
numerous (over 400) free auditory clubs.

The eighteen strong tentacles, which alternate with the eighteen collar lobes and are
inserted at their basis in the coronal furrow, were generally curved upwards in the
specimen examined, as often happens in the Narcomedusse (fig. 1). They are cylindrical,
somewhat thicker towards the base, thinner towards the point, and nearly as long as
the diameter of the umbrella. The endodermal axis is composed of a single row of coin-
shaped chordal cells (a millimeter broad), in which the nuclei form a central chain
(fig. 11). A visible layer of longitudinal muscular fibres (m) lies outside the strong-
supporting plate (fig. 10,2). The spheroidal nematocysts (n) in the exodermal epithelium
are chiefly accumulated in the abaxial side of the tentacles, most thickly at the point.
A thick, almost closed, urticating ring, which has only a break internally on the axial
side (figs. 2, 3, nb) is placed at the insertion of the tentacles. Inside this the tentacles
run into the pointed conical roots (tr) which pass centripetally into the gelatinous sub-
stance of the umbrella,

EEPORT ON THE DEEP-SEA MEDUSiE. 39

The auditory clubs in Pegantha pantheon are very numerous, 400 to 450, 20 to 25 upon
each of the collar lobes (figs. 1, 2). They are regularly distributed along the bow-shaped
umbrella margin, and run inwards immediately into the auditory clasps. The otoporpae
or auditory clasps (figs. 2, 4, oo) are all equal in length, nearly three times as long as
the auditory clubs ; their axis converges more or less towards the middle point of the lobe
base (fig. 2). Their inner end is thickened and rounded like a club ; their exodermal epi-
thelium contains many larger and smaller thread-cells. The auditory clubs themselves are
larger than in most other Narcomedusse, and contain an axis of three to five, usually four,
large endodermal cells, each enclosing a crystal. The proximal otolite (at the thin end of
the auditory club) is the smallest, the distal otolite (in the club-shaped rounded end) the
largest, and between these one or two medium-sized crystals (in the middle endodermal
cells). The auditory club is surrounded by a stiff bunch of auditory hairs (oh), which
run out from the auditory pad (op). After treatment with acetic acid and carmine, a
nucleus coloured red (perhaps the original cell nucleus enclosed by the formation of the
otolite) was visible in the centre of the otolite (fig. 12, ol). The manner in which the
auditory clubs and their auditory clasps are disposed upon the umbrella margin, and their
relations to the adjacent organs will be best understood by a comparative study of
figs. 2 and 4 in Plate XL and figs. 7 and 12 in Plate XII. In fig. 12 especially it
is clear how the auditory club rises on a thin stalk from the conical auditory pad (op)
of the urticating ring (»c), and how the fulcral lamella (z) between the two nerve
rings (?ic) passes through to the base of the auditory club, and thence into its sup-
porting plate.

The gastrovascular system (PI. XL figs. 1,3; PI. XII. figs. 7, 9, 12) is, on the whole,
the same as that already described in Polycolpa forskalii (comp. above, p. 34, and
PI. XL figs. 1, 3, 6, 8). The stomach also forms a wide, flat, circular pouch, occupying
the whole lower side of the umbrella lens (fig. 7, gc). Whilst the slightly convex
lower surface of the latter forms the cover of the gastral cavity, its bottom is formed
by the very muscular and extensible central part of the subumbrella, which is laid in a
large number of radial folds (fig. 9). A short, wide oesophagus hangs down in the
middle of the folds, its ample oral opening showing a swollen, thickened oral margin
(fig. 7, qg). The peripheric part of the gastrovascular system is formed by the festoon
canal, which is composed, in this species, of eighteen separate lobe canals (comp. above,
p. 35). The latter run on the inner side of the urticating ring along the margin of the
oval collar lobes, and open at their bases immediately into the periphery of the gastral
cavity (fig. 3, go). The two openings take in between them the stalk of the genital
sacs, which hangs in the relative lobe cavity (fig. 3, oc). The ribbon-shaped flattened
canals are nearly one-eighth as broad as the greatest breadth of the lobes. Their sub-
umbral endodermal epithelium (fig. 12, div) is composed, as usual, of very high narrow
cylindrical cells, whilst that of their umbral wall (cho) consists of many flat, flagellate

40 THE VOYAGE OF H.M.S. CHALLENGER.

cells. A low cudodermal fold, like that in Pedis (PL VI. fig. 12, yc), but less strongly-
developed, rises at the distal margin of the festoon canal, projecting freely into its
lumen (PL XII. fig. 12, yc).

In contrast to the preceding Polycolpa, in which the reproductive glands appear in
their simplest form as a circular closed girdle in the lower wall of the stomach,
Pegantha shows us the most widely differentiated and highly developed form of the
genitalia. Here the originally simple and connected genital girdle is divided into a
circle of separate reproductive sacs, hanging freely in the periphery of the gastral
cavity. Each lobe cavity of the umbrella collar receives one genital pouch, which is
surrounded and protected by the concave subumbral surface of the collar lobes (PL XI.
fig. 3 ; PL XII. figs. 7, 9). The genera of the Peganthidse, Polyxenia and Pegasia
represent connective intermediate forms between the two extremes Polycolpa and
Pegantha, so that we have here the division of the simple subgastral reproductive
girdle into a circle of separate pouches, shown in four different phylogenetic stages
(comp. my System der Medusen, 1879, pp. 327-332). The cavities of the isolated
reproductive sacs of Pegantha pantheon (fig. 5, sc) communicate with the periphery
of the gastral cavity (figs. 7, 9, sc). Each of the eighteen genitalia has the shape of a
thick roundish leaf, with the two edges turned towards the edges of the collar lobes,
whilst the upper surface is delicately twisted or folded (fig. 3, s). The transverse section
(figs. 5, 6) shows that the gastral endodermal epithelium of the sac cavity (sd) is com-
posed of high cylindrical cells, and divided by a strong supporting plate from the mass
of the spermatozoa (sm). On the other hand, the latter is in continuous connection
with the gastral ectodermal epithelium of the subumbrella, from which it originates.
Under stronger magnifying power, we find the same condition here which Hertwig
described (1878) in Cunina lativentris. The superficial ectodermal layer of cells
(figs. 5, 6, sio), which forms the subumbral cover of the testes, sends out supporting-
fibres containing nuclei (zs) into the subepithelial layer of cells lying beneath it. The
larger cells of this layer (sm) usually lie inwards, touch the endodermal fulcral plate
(2), and must be regarded as " mother cells of the spermatozoa," whilst the smaller cells,
which usually lie outwards, form spermatozoa already ripe (fig. 6, sz).

Family, jEginid^;, Gegenbaur, 1856.

JSginid^:, Hasckel, System der Medusen, 1879, p. 334, taf. xix. fig. 8, 9; taf. xx. fig. 11-16.

Narcomedusse with a marginal canal communicating- immediately with the stomach by
double peronial canals, with internemal gastral pouches (which have arisen from the
distal lobe pouches of radial canals through retrograde formation), without otoporpse or
auditory clasps at the basis of the auditory clubs.

REPORT ON THE DEEP-SEA MEDUSAE. 41

Sub-family, ^Eginukid^, Hseckel, 1879.
iEginidse with eight peronial double canals (four perradial and four interradial).

sEginura,1 Hseckel, 1879.

iEginidse having eight peronial double canals and eight tentacles (four perradial
and four interradial), and also sixteen intern emal reproductive pouches, alternating in
pairs with the eight tentacles.

The genus JEginura shows twice as many tentacles and reproductive pouches as
the well-known primitive genus of the iEginidas, ^Egina. The ^Eginura myosura,
which I am about to describe, is the only species of this genus, and also the only
species of iEginidag of the Challenger collection, of which I can give a satisfactory
account. A second species of this family, much larger, and of more complicated struc-
ture, which I placed in my System, 1879, as sEginorhodus rosarius (p. 345), proved
on closer examination too much destroyed and badly preserved for any satisfactory
description. I am even dubious if the diagnosis of the species taken from this fragment
be correct.

JEginura myosura, Hseckel (Pis. XIII., XIV.).

JEginura myosura, Hasckel, System der Medusen, 1879, p. 343, taf. xix. figs. 8, 9.

Umbrella cap-shaped, twice as broad as high. Mouth quadrate or four-lobed, with
cylindrical oesophagus half as long as the radius of the umbrella. Sixteen almost
rectangular reproductive pouches, the two medial of each quadrant smaller than the two
lateral. Eight tentacles alternately different ; the four perradial larger, and inserted
higher than the four interradial ; the former nearly double the length of the radius of
the umbrella, the latter the same length. Horizontal diameter, 30 mm. ; vertical
diameter, 15 mm.

Habitat. — Indian Ocean, south of Australia. I found a male specimen of this species
somewhat damaged, but still pretty well preserved, in a bottle of the Challenger collec-
tion, containing numerous Phseodaria (Atdosphera, Ccelodendrum, &c.) from Station
159. Lat. 47° 25' S., long. 130° 32' E. 10th March 1874. Depth (apparently) 2150
fathoms. In the System der Medusen (p. 343), "Weber" was given by an oversight
instead of " Challenger."

The umbrella (PI. XIII. figs. 1, 2 ; PL XIV. fig. 11) has the shape of a flat cap or
biretta ; the upper surface flattened nearly horizontally (slightly depressed in the
middle), whilst the side walls stand almost vertically (slightly widened below). The

1 JEginura, nonien proprium.

(ZOOL. CHALL. EXP. PART XII. 1881.) M G

42 THE VOYAGE OF H.M.S. CHALLENGER.

largest horizontal diameter near the umbrella margin amounted to 30 mm., double the
vertical height of the umbrella (15 mm.). As the umbrella margin in the specimen
examined was strongly contracted, the height in the living animal must be proportion-
ately greater (20 mm. or more). The umbrella, seen from the upper or lower surface,
appears distinctly octagonal, as the eight principal radia (with peronia and tentacles)
project more strongly outwards than the eight interlying side walls (fig. 2), so that the
umbrella of the dead Medusa has really the shape of a short, regularly octagonal prism.
The gelatinous substance of the umbrella is tolerably soft (as in all true iEginidse), not
so firm as in the Cunanthidae and Peganthidse, though there, as here, it is traversed by
numerous elastic fibres. The gelatinous umbrella is very thick throughout the flattened
apical surface (equal to one-third of the height of the umbrella) but very thin (and
decreasing proportionately below) on the thin lateral walls (fig. 11, ug).

The exumbrella is flat, without any special characteristic, and only traversed by eight
shallow peronial furrows (fig. 7, es) ; these run vertically from the insertion of the tentacles
to the umbrella margin, and are connected by thin " peronial plates " with the peronia or
"umbrella clasps" lying beneath them (em). The eight peronial plates ("laminae peroni-
ales," figs. 7, em; 12, em) consist of a double layer of the exodermal flat epithelium of the
exumbrella, and originate from the two gelatinous walls of the umbrella, which limit the
open peronial groove laterally in the Cunanthidae (PI. IX. fig. 5, es), but lie above the
grooves with their edges fused together in the iEginidae. The peronium in the ZEginidae
is therefore completely enclosed by the gelatinous substance of the umbrella on the
abaxial side, and by the subumbrella on the axial side, whilst in the Cunanthidae the
abaxial side of the peronium lies free at the bottom of the open peronial groove (PL IX.
fig. 5, en). In JEginara the distal end of the peronium joins that of the peronial plate
under the umbrella margin ; both pass continuously into the marginal urticating ring
(PI. XIII. figs. 1, 2, 4, nc ; PL XIV. fig. 11, nc).

The umbrella cavity and the subumbrella lining it do not present in JEginura, any
more than in the other iEginidae, any of the striking peculiarities which distinguish the
two families of the Narcornedusse, the Cunanthidae, and Peganthidae ; the conditions do not
differ essentially from those usual in the Craspedotse. Hence it comes that the umbrella
collar is not divided into separate lobes by deep peronial incisions, and the margin of the
umbrella is therefore almost entire. The peculiar lobe cavities of the Cunanthidae (PL IX.
fig. 6, nl) and the Peganthidse (PL XII. fig. 7, nl) are consecpiently wanting. In
JEginura the umbrella cavity is more a simple cylindrical, or almost octagonal, hollow
space, with the oesophagus hanging in its axis, whose horizontal roof is formed by the
subumbral bottom of the stomach (PL XIV. fig. 11, gw), whilst it opens wide below, and
is limited laterally by the vertical side walls of the subumbrella. The latter has an
unbroken broad layer of circular muscular fibres, which is divided by the eight peronia
into eight quadrangular plates, but not cut through by it (comp. figs. 7, 11, 12, mw).

REPORT ON THE DEEP-SEA MEDUSA. 43

The velum (v) in JEginura, as in the other iEginidse, is tolerably broad and strong,
though not so thick and muscular as in the Cunanthidse and Peganthidse. Nor does it
form the peculiar lateral points which project vertically into the deep peronial incisions
of the umbrella margin, between each two collar lobes, as in the last-named families.
As the collar lobes in the iEginidse are not divided, but united below as far as the
umbrella margin by the peronial plates, the velum is nearly of equal thickness through-
out, and only slightly broader at the eight principal points where the peronia rise out
from the eight corners of the umbrella margin (PL XIII. figs. 2,v; 4, r).

The umbrella margin (fig. 2) appears regularly octagonal from the slight curvings
above mentioned at the principal points where the distal ends of the peronia and peronial
plates run out from the urticating ring of the umbrella margin. Its cylindrical urticat-
ing ring is limited inside by the velum, outside by the distal margin of the exumbrella,
above by the annular canal, and below by the double nerve ring. The sixteen subradial
auditory clubs are placed on the abaxial external surface of the urticating ring, between
the lower margin of the gelatinous umbrella and the upper margin of the velum, whilst
the eight tentacles are inserted much further up on the exumbrella, at the height of the
covering of the stomach.

The sixteen auditory clubs ("cordyli," PI. XIII. figs. 1-3, ok) are placed strictly sub-
radially, i.e., they lie in the sixteen radial meridian planes, exactly in the middle between
the eight adradial planes of third order, and the eight principal planes, in which the four
interradial tentacles (second order) and the four perradial tentacles (first order) are placed.
The auditory clubs are placed upon a hemispherical auditory pad (fig. 3, op), a wart-like
swelling of the urticating ring, which apparently conceals a "ganglion acusticum;" at least
the ganglion cells of the dorsal nerve ring are much more thickly accumulated there (fig.
10), and are connected immediately by numerous fine nervous fibrillae, with the high
cylindrical sense cells covering the auditory pad. A thick bunch of very long, fine
auditory hairs radiate from the pad (fig. 3, oh). The auditory club, which is very thin
at the base and considerably enlarged at the distal end, is placed in the axis of the conical
bunch. Its endodermal axis consists of 5-6 chordal cells, of which the 2-3 proximal
are very small and without otolites, but the 3-4 distal very large, and enclose otolites.
The largest otolite in the terminal endoderm cell is sometimes nearly half as long as
the whole auditory club. The ectodermal covering of the latter is divided from the
endodermal axis by a delicate fulcral lamella (2), and consists of flat sense cells

(%• 3, q).

The eight tentacles are inserted high up on the exumbral surface, far from the
umbrella margin, and only connected immediately wTith it by the eight long, strong
peronia. A conical tentacle root (fig. 11, br) runs inwards from the point of insertion
(at the proximal end of the peronial furrow), horizontally and centripetally, into the
gelatinous substance of the umbrella, whilst the peronium (em) goes downward almost at

44 THE VOYAGE OF H.M.S. CHALLENGER.

a right angle to the umbrella margin. All eight tentacles have the same form and
structure ; the four primary perraclial tentacles are, however, twice as long as the four
secondary interradial, the former are also inserted somewhat higher, and the clasps of the
latter are consecmently somewhat shorter. The four perraclial tentacles are somewhat
longer than the largest diameter of the umbrella, the four interradial only about half so
long. The free cylindrical tentacle filament (fig. 5, longitudinal section ; fig. 6, seen
from the outside) is more than a millimeter thick at the base, decreases towards the
point like an awl, and is shaped like a mouse's tail ("myosura"). The solid axis
resembles a rouleau of coin, and consists of a single row of discoid chorda! cells whose
nuclei be in the centre, one behind the other (fig. 5, ym ; comp. also PL XII. figs. 10, 11).
The elastic structureless supporting plate enclosing this column of chordal cells (2), is
covered by a layer of longitudinal muscular fibres (fig. 6, mt), above and outside which
lies the single layered epithelium of the ectoderm (d). The spheroidal nematocysts (m) in
the exoderm lie thickly together on the dorsal (abaxial) side of the tentacles, and form a
raised urticating band (fig. 6, m), whilst they are only scantily distributed and of smaller
size on the other sides of the tentacles.

The peronia or " umbrella clasps," which serve to connect the base of the tentacle
with the urticating ring of the umbrella margin, are eight thick urticating streaks,
gradually increasing in breadth from the top to the bottom (figs. 1, 2, 4, en). They
appear egg-shaped in transverse section (figs. 7, en; 12, en), and under higher magni-
fying power they prove to be composed of the peculiar " peronial tissue " or " urticating
skeletal tissue," which is the most important element in the urticating ring, and in the
peronia and otoporpse of the Narcornedusse. This tissue (fig. 12, en) consists of com-
pacted exodermal thread cells, varying greatly in shape and size. The roundish thread
cells containing nematocysts enclose a long urticating thread, wound thickly and spirally ;
they have very thick walls, and are partly much larger (three to four times as large) than
the ordinary largest nematocysts of the tentacles. These nematocysts are plainly incapable
of throwing out their threads, but only serve with their thickened wall as firm " support-
ing cells." The inner axial side of the peronia is then closed on the exodermal epithe-
liuin of the subumbrella (figs. 7, 12, qw) ; also on the peronial canals, touching them
laterally (cts) by a thick supporting plate, whilst its outer abaxial side touches the
gelatinous substance of the umbrella (wg).

The tentacle roots (fig. 11, tr) are, as usual, conical, being a centripetal prolongation of
the tentacle axis, consisting of a few large chordal cells of the endoderm, and having their
points directed centripetally. They are covered by a structureless supporting plate, but
have no exodermal epithebum. Their dorsal and their lateral surfaces are enclosed in
the gelatinous substance of the umbrella, whUst their ventral surface Hes immediately on
the cover of the stomach (or the dorsal gastral wall), which it serves at the same time to
support.

REPORT ON THE DEEP-SEA MEDUS.E. 45

In JEginura, as in the other true iEginidse, the gastrovascular system (PL XIII. figs. 1,
2, 4, 7 ; PI. XIV. figs. 8, 11, 12) consists of two different principal parts, corresponding
to the two principal parts of the umbrella, and separated by its coronal furrow (ec). The
central gastral cavity with the oesophagus and oral opening lies on the subumbral side
of the central lens of the umbrella, whdst on the subumbral side of the peripheric corona
of the umbrella there is a circle composed of sixteen interuemal reproductive pouches and
eight peronial double canals, connected with the umbrella margin by an octagonal
marginal canal, along with which it forms the " festoon canal." The central gastral
cavity is fiat and wide and regularly octagonal in outline (corresponding to the eight
tentacle roots and peronial furrows). The cover of the stomach or the upper umbra 1
wall is formed by the flat or slightly convex gastral surface of the gelatinous umbrella
lens, into which the adjacent tentacle roots (tr) project centripetally as eight supporting-
edges. The bottom of the stomach or the lower subumbral wall, consists of a thick layer
of circular muscular fibres, immediately connected on their lower surface with the
ectodermal epithelium of the subumbrella (w), but separated on their upper surface by a
thick supporting plate (s) from the high endodermal epithelium of the stomach. A
cylindrical oesophagus (gt) hangs from the middle ; it is nearly half as long as the whole
radius of the umbrella, probably considerably longer in the living animal. The
oesophagus is nearly as broad as long, and quack-angularly prismatic towards the lower
end, where the four interradial longitudinal furrows appear, which divide the four broad
bordering oral lobes (fig. 8).

The coronal intestine, which projects from the periphery of the central principal
intestine, is composed of a circle of sixteen reproductive pouches, alternating in pairs
with eight peronial double canals, and arises from the octagonal marginal canal
along with which it forms the festoon canal. This festoon canal (" canalis festivus ") is
homologous with the festoon canal already described in Cunarcha, Polycolpa, and
Pegantha, and really consists of eight interuemal " lobe canals," which edge the margin
of the eight cmaclrangular collar lobes of the corona of the umbrella. In AZginura,
however, the lateral margins of these " collar lobes " are fused into the eight peronial
furrows (es), so that each lobe canal is divided into a horizontal middle part (an octant
of the marginal canal) and two vertical side limbs (the two inverted halves of two
peronial double canals). At first sight a simple " circular canal " appears to exist upon
the umbrella margin (PL XIII. figs. 1, 2, 4), which is connected with the stomach by
the eight simple broad "radial canals" (as in many Craspedotse, such as the Pectyllicke,
Pis. III. -VIII.). In the transverse sections (figs. 7, 12), however, we see at once
and indubitably that the eight broad, apparently simple "radial canals" consist of
two isolated "peronial canals" (ck) fully separated by the peronium. Each of the
two adjacent peronial canals or clasp-canals open independently above (beside the
insertion of the tentacles) into the periphery of the stomach, whilst it turns almost at

46 THE VOYAGE OF H.M.S. CHALLENGER.

right angles into the corresponding pieces of the marginal canal. The marginal canal
("canalis marginalia," cm) which runs along the proximal side of the urticating ring, is not
however the usual marginal " circular canal " of the Medusae margin, but consists of
eight completely distinct pieces, separated from each other by the distal ends of the
peronia. Each of these independent " octants of the marginal canal " runs at the two
ends into a peronial canal, the two branches thus forming a horseshoe-shaped "lobe
canal." Each lobe canal opens with two separate mouths into the gastral cavity beside
the base of insertion of each two tentacles. The two peronial canals of each double
canal and their two gastral openings (at both sides of a tentacle) therefore belong to two
different "lobe canals." The eight lobe canals form collectively the eight-lobed
" festoon canal," and this is phylogentically only a peculiar modification of the
primarily simple "circular" canal, caused by the dorsal change of position of the
tentacles and the formation of peronia connected with it.

The sixteen subradial reproductive pouches of JEginura show essentially the same
formation already described by Mertens in jEginopsis laurentii (1838, loc. cit. PI. VI.).
They are quadrangular, almost rectangular, and distributed in such a way that a large
and a small pouch is placed on each of the eight collar lobes (PI. XIII. fig. 1, 2). The
pouches, consequently, lie in internemal pairs, a pair between each two tentacles and
peronia. It appears, however, on closer inspection that, as in ^ginopsis laurentii, all
the sixteen pouches actually belong to four primary groups. Two smaller pouches are
placed on both sides of the four larger perradial tentacles, and two larger pouches on both
sides of the four smaller interradial tentacles. If the whole umbrella be divided into
four quadrants, whose middle lines form the four perradial peronia and the border lines
the four interradial peronia, a group of pouches consisting of two small medial pouches and
two large lateral pouches falls in each of the quadrants. The same condition is shown,
if we suppose each of the eight lobe pouches of Cunarcha already described (PI. IX.
figs. 2-4, bl) divided by a centripetal incision of their distal margin into two pouches
of unequal size, and the four proximal (perradial gastral pouches), formed by the bifur-
cation of the eight lobe pouches, to have undergone retrograde formation. It is then clear
that each group of four associated reproductive pouches belonging together in JEginura,
is simply the double bifurcated distal part of a perradial gastral pouch, whose undivided
proximal part has undergone retrograde formation (or become part of the central stomach).

In fact, it is only by such morphological comparison that we can understand phylo-
genetically the remarkable and varied conditions of vascular formation in the iEginidas.
The peculiar, apparently isolated, gastrovascular system of the iEginida?, is, therefore,
naturally derived from that of the Cunoctonidse, from those Cunanthidfe {Cunarcha,
Cunoctona, Cunissa) in which each radial canal (or each "pernemal gastral pouch") is
cleft at the distal margin into two csecal lobe pouches. If these paired lobe pouches
become larger, and the undivided proximal piece of the pernemal gastral pouch under-

11EF0RT ON THE DEEP-SEA MEDUSAE. 47

goes retrograde formation at the same time, the former originate the characteristic
internemal gastral pouches of the iEginidae, which were primarily placed in pairs between
every two tentacles (as in the tetranemal /Egina, System, 1879, p. 337, taf. xx.). The
two pouches which have a tentacle between them are therefore the distal halves belong-
ing to a former pernemal gastral pouch, i.e., of a radial canal at the end of which each
tentacle was originally placed. But the two pouches lying between every two primary
tentacles are opposite distal halves of two adjacent radial canals. This view is justified
by the fact, that in all the older and simpler forms of the iEginidse two gastral pouches
are always placed between every two tentacles. In our JEginura (as in AZginopsis) each
of the eight lobe pouches is divided a second time. The peculiar formation of the
festoon canals of the iEginidse can only be explained in this way. It shows essentially
the same conditions as in the Cunanthidge. Here as there, the originally simple
circular canal is divided into the same number of separate arches or " lobe canals " as
there are umbrella lobes, and each lobe canal opens with two mouths beside the base
of two neighbouring canals. But whilst in the Cunanthidse the opening of the lobe canal
is found in the middle of the distal margin, in the iEginidse it occurs immediately in
the periphery of the stomach. In the former the undivided proximal part or principal
part of the radial canals (or of the pernemal gastral pouches) has entirely disappeared, and
the internemal lobe pouches only are left (as remains of the divided distal part). The
inverted halves of every two adjacent lobe canals are also connected with a " double
canal" or double " peronial canal." In the iEginidse, as the proximal half of the
umbrella margin has retrograded, and the distal half become proportionally more strongly
developed, the double canal appears very much prolonged, and has the deceptive
appearance of " a simple radial canal opening into the periphery of the stomach
between every two internemal gastral pouches." Thus very simple and clear homologies
exist between formations apparently very different, as I have already shown in my
System der Medusen, 1879, pp. 305, 306, &c.

The specimen of ^Eginura myosura was a male, and its sixteen testes (the sixteen
"internemal gastral pouches") contained masses of ripe spermatoza. They did not however,
fill up the cavity of the pouches, but were placed on the outside of its subumbral wall.
In transverse sections the internal side of the subumbral wall showed the same high
cylindrical epithelium as that of the peronial canals (figs. 7, 12), whilst the endodermal
epithelium of the opposite umbral wall consisted, in both cases, of a thin layer of flat
plate cells covering the gelatinous substance of the umbrella. The spermarium,
on the contrary, lies like a thick plate immediately under the exoderm epithelium
of the subumbrella from which it originates, and is divided from the high cylindrical
epithelium of the endoderm by a distinct supporting plate. In jEginura as in
Pegantha (p. 34, PI. XL figs. 5, 6), the subumbral ectoderm sends out supporting fibres
containing nuclei into the spermarium which lies under it, and is derived from it. Here,

Js THE VOYAGE OF H.M.S. CHALLENGER.

therefore, as in all Craspedotee, the sexual products originate from the ectoderm,
whilst the reverse is the case in the Acraspedotse, where, in both sexes, they are
formed by the endoderm. The ripe spermatozoa and the ripe ova are therefore thrown
outside immediately in the Craspedotse or " Cryptocarpse," whilst in the Acraspedse
or " Phanerocarpge " they first pass into the gastrovascular system, and are then ejected
by the oral opening ; the former are therefore properly " Ectocarpae," the latter
"Entocarpa?" (Hertwig, 1879). When Eschscholtz, the meritorious founder of the
system of the Medusa?, 1829, distinguished the two principal divisions of this class as
" Cryptocarpse " and " Phanerocarpse " according to the different formation of their repro-
ductive organs, he expressed prophetically an important difference, whose peculiar
character was first more accurately recognised fifty years later.

Class II. ACRASPEDSE, Gegenbaur, 1856.

Phanerocarpge, Eschscholtz, 1829. Steganophtaljle, Forhes, 1848.

ScyphomeduszE, Ray-Lankester, 1877. Phacellotje, Ha;ckel, 1878.

Medusae with gastral filaments or phacellas ; with endodermal genitalia (or sexual
products from the internal germinal layer) ; without true velum (often with velarium) ;
with true marginal lobes of the umbrella; without double centralised nerve ring. Phylo-
genetic descent (probably universal) and ontogenetic descent (at present still in the
majority) derived from scyphopolyps with gastral filaments or from scyphostoma.
Ontogenesis usually alternation of generations (in the form of strobilogenesis) often
connected with metamorphosis. The sexual acraspede generation is formed by terminal
gemmation from the asexual scyphostoma generation.

Order V. STAUROMEDUS^, Haeckel, 1877.

Acraspedaa without sense clubs, with simple tentacles or marginal anchors (adhesive
tentacle rudiments). Originally eight principal tentacles (sometimes rudimentary).

Besides these eight principal tentacles there are often small secondary tentacles
(usually in bunches upon eight adradial marginal lobes). Stomach having four wide
perradial gastral pouches, which are separated by four narrow interradial septa or fused
selvages, and connected on the umbrella margin by a circular sinus. Genitalia,
four interradial horseshoe-shaped swellings or four pair of adradial swellings, which are
developed in the subumbral wall of the gastral pouches from their endoderm, and project
wholly or partially into their cavity.

REPORT ON THE DEEP-SEA MEDUSA. 49

Family, Tesserid^e, Hseckel, 1877.

TESSERiDiE, Hreckel, System der Medusen, 1879, p. 371, taf. xxi.

Stauromedusae with simple, undivided umbrella margin, without hollow marginal
lobes or " arms." Eight principal tentacles (four perradial and four interradial) always
present, not transformed into marginal anchors or sense clubs ; besides these, sometimes
numerous secondary tentacles. Coronal muscle of the umbrella margin circular, not
divided into eight isolated marginal muscles. Either an apical process or an umbrella
peduncle on the apex of the umbrella.

Sub-family, Tesseranthid.e, Haackel, 1879.

Free-swimming Tesseridaa, without a stalk, but with an apical process on the cone of
the umbrella ; with simple solid tentacles without terminal urticating knob.

Tesserantha,1 Hseckel, 1879.

Tesseridaa, without peduncle, with an apical process and with sixteen simple solid
tentacles without terminal urticating knob (four perradial, four interradial and eight
adradial). The genus Tesserantha is one of the simplest and oldest Medusae forms of that
important family the Tesseridse, which are to be regarded as the general ancestral
group of all Acraspedaa. This primitive Acraspeda form is essentially merely a Scy-
phostoma with sixteen tentacles which, in adapting itself to a free-swimming mode of life,
changed its oral disc into a subumbrella, and its basal peduncle into an apical process,
divided the peripheric gastral space into four radial pouches by four interradial fused
knobs, and became sexually mature in this form. Tesserantha is distinguished from the
octonemal closely related Tessera by the addition of eight new adradial tentacles (of the
third order) to the eight principal tentacles (four perradial and four interradial). More-
over, whilst in Tessera only four simple gastral filaments run out from the four septal
knobs, as terminal free processes of the four interradial tamiola, the septal knobs in
Tesserantha are beset with a double row of filaments throughout the greater part of their
iength (the proximal basal part alone excepted). In this and other respects, namely in the
formation of four perradial mesogonial folds and four interradial funnel cavities alternating
with these, Tesserantha comes nearer De'pastrella, and therefore forms an interesting-
transition gradation between Tessera and Depastrella. At present there is only one
known species of this genus, the deep-sea Medusa described below.

1 Tesserantha, four-sided flower.

(ZOOL. CIIALL. EXP. PART XII. 1881.) M 7

50 THE VOYAGE OF H.M.S. CHALLENGER.

Tesserantha connectens, Hasckel (PI. XV.).

Tesserantha connectens, Hreckel, 1879, System der Medusen, p. 375, No. 402.

Umbrella helmet-shaped, one and a half times as high as broad, with conical apical
process and peduncle canal on the top. Exumbrella with eight longer and eight shorter
exumbral urticating ribs. Four double rows of gastral filaments along the four inter-
radial tseniola inside the central stomach. (Esophagus quadrangularly prismatic, half as
long as the height of the umbrella. Oral opening with four short, frilled, oral lobes.
Eight simple horse-shoe-shaped genitalia enclosing the small septal nodes in the con-
cavity of the arch. The eight principal tentacles (four perradial, four interradial) of ecpual
length, nearly as long as the height of the umbrella ; the eight secondary (adradial)
tentacles only half as long. Horizonal diameter of the umbrella, 6 mm. ; vertical
diameter, 9 mm.

Habitat. — South-east part of the Pacific Ocean, not far from the island of Juan
Fernandez, lat. 33° 31' S., long. 74° 43' W.; Station 299. 14th December 1875.
Depth, 2160 fathoms.

The umbrella (figs. 1-3) is highly vaulted, bell or helmet-shaped, rather constricted
beneath at the opening, just above the umbrella margin, and furnished above with
a pointed, conical, apical process which is nearly a third as long as the whole height of
the umbrella. It is about half as large again as the greatest horizontal diameter of
the umbrella above the umbrella margin. The exumbrella is distinguished by eight
projecting, strong, urticating ribs, four perradial and four interradial (figs. 1, er; 6, er).
These are sharp corners of the outer surface of the umbrella, almost triangular in
transverse section, which are armed with a broad streak of pigment cells and thread cells,
and stretch uninterruptedly from the point of the umbrella cone to the eight ocelli of the
umbrella margin, from which they pass on to the dorsal surface of the eight principal
tentacles. Eight secondary incomplete longitudinal ribs alternate with the eight prin-
cipal complete longitudinal ribs of the exumbrella ; these are much narrower and shorter
and only run from the bases of insertion of the eight adradial tentacles to half the height
of the umbrella (fig. 1 , er).

The umbrella margin is somewhat contracted by a circular marginal stricture, and is
beset with sixteen tentacles, between which the gelatinous substance of the umbrella
projects a little in the form of short, roundish, solid, gelatinous lobes (figs. 1, 4, I). The
eight principal tentacles (four perradial and four interradial) are nearly as long as the
height of the umbrella, whilst the eight adradial or secondary tentacles alternating with
them, are only half as long. These also want the black roundish eye-spot (" ocellus,"
fig. 1, oc) which is found at the base of the eight principal tentacles. These ocelli
consist of accumulations of black grains of pigment in the ectoderm of the tentacle
basis. All the sixteen tentacles are solid cylindrical filaments, gradually becoming

EEPORT ON THE DEEP-SEA MEDUSAE. 51

thinner towards the pointed distal end. Their structure resembles that of the oral
styles of the Margelidse (p. 1, PI. I. fig. 5) and of the solid tentacles of the
Peganthidse (p. 30, figs. 10, 11, &c.) already described. Each tentacle, therefore,
consists of four different layers: (1) a solid cylindrical endodermal axis, formed of a single
row of clear coin-shaped chordal cells; (2) a thin but firm and very elastic fulcral plate ;
(3) a thin muscular plate composed of parallel longitudinal fibres ; (4) an exodermal
epithelium, bearing partly thread cells, partly pigment cells. The latter contain grains
of blackish pigment, and are chiefly found on the abaxial or dorsal side of the tentacle,
where they form a black longitudinal streak which represents the direct process of the
ocelli and the exumbral ribs of pigment. The tentacles, with their basal ocelli, are the
only organs of sense found in the Tesserantha, as in all Stauroniedusae ; special auditory
clubs, like those of the other Acraspedse, are not present.

The deep cavity of the umbrella (subumbrella) consists of a lower simple cavity of the
corona of the umbrella, whose vertical axis is occupied by the oesophagus (fig. 2, at), and of
an upper quadrilocular part divided by four mesenteries into four conical funnel cavities
(fig. 6, ii). These mesenteries or mesogonia (fig. 2, wr) are four thin perradial mem-
branes, which stretch vertically between the four perradial angles of the base of the
stomach and the middle line of the four radial pouches. They serve principally to attach
the oesophagus, are cut out like a crescent at the lower free edge, and pass immediately
into the tissue of the subumbrella at the upper, rather thinned, basal margin. The
mesenteries must be regarded essentially as folds of the subumbrella, whose structure
they share. We find them again in a similar form in the Charybdeidse, Tiaridae, and
Pectyllidse (PI. IV. fig. 3, ivr ; PI. VIII. fig. 9, wr). The four interradial funnel
cavities (" infundibula subumbralia," fig. 6, ii), which are divided by the four mesenteric
folds, are conical sacs, opening below into the umbrella cavity, but projecting more or
less with their caeca! point into the central gastral cavity ; their aboral extension could
not be exactly defined ; they perhaps extend as far as the tseniola are set with filaments,
to the beginning of the basal stomach.

The muscles of the subumbrella are formed by two different systems, which are found
more or less modified in all Acraspedse ; a distal system of circular muscular fibres and a
proximal system of radial muscular fibres. The first form the typical coronal muscle
("museums coronarius," figs. 2-4, me), a broad octagonal ring on the umbrella margin,
whose eight angles are defined by the bases of the eight adradial tentacles. The system
of radial or longitudinal muscles is composed of eight triangular deltoid muscles, whose
broad base rests on the proximal margin of the coronal muscle. The four perradial
deltoid muscles (figs. 3, 4, md') are narrower and longer, and pass above into the
mesenteric folds. The four interradial deltoid muscles (figs. 3, 4, md") are broader and
shorter, and their truncated point is inserted at the four septal nodes (kn).

In Tesserantha, as in all Acraspedse, the " gastrovascular " system consists of two

52 THE VOYAGE OF H.M.S. CHALLENGER.

principal parts, the central principal intestine and the peripheric coronal intestine. The
central part or the axial principal intestine (" gaster principalis ") communicates with the
peripheric coronal intestine by the four perradial openings (" ostia gastralia"), and is
divided into three different sections, the basal, the central, and the oral stomach. The
aboral basal stomach or peduncle canal (" gaster basalis,"</6), which may also be called the
" apical canal," is a narrow, almost cylindrical, hollow space occupying the entire cone of
the umbrella, in whose point it ends csecally above, whilst below it opens by the pylorus
(gy) into the central stomach. Four longitudinal gelatinous selvages, the important
interradial tamiola, project from the inner surface into its hollow space, and, as in
the closely-allied Lucernaridaa, traverse the entire length of the hollow basal umbrella
peduncle (figs. 2, 3, 8, ft). The peripheric part of the basal stomach is thus divided into
four perradial grooves (figs. 3, 8, gb).

The central stomach (" gaster centralis," gc) has, on the whole, a spheroidal or almost
quadrangularly pyramidal form, which, however, is complicated by the four interradial
exodermal funnel cavities (ii) sinking down into it from above. The distal processes of
the four tseniola, each of which bears two rows of gastral filaments (ft) inside in the central
stomach, run as projecting selvages on the endodermal gastral surface of the funnel
cavities. The central stomach opens above by the " porta pylorica " (gy) into the basal
stomach, below in the centre by the "porta palatina" (gp) into the oral stomach, and round
by the four cleft-shaped gastral openings into the coronal intestine. The gastral open-
ings (fig. 6, go) are narrow, almost horizontal clefts, divided from one another by the
four interradial septal nodes ("nodi cathammales," kn), these important points of fusion
at which the umbral and the subumbral wall of the gastral space have grown together.
That this is really a fused plate is plain from the fact that an endodermal layer of
epithelium — " endodermal lamella " or cathammal plate — runs in the middle through the
cartilaginous-like gelatinous mass of the septal node.

The oral stomach or oesophagus ("proboscis," go) is formed of a quadrangularly
prismatic tube, nearly equal in length to the breadth of the umbrella (figs. 1, 2, at). It is
four times as long as broad, and has four projecting perradial angles which run into the
four mesenteries above, whilst the external surface is depressed like a groove between
them (fig. 6, a). The oral opening is surrounded by an undulating oral margin
crowded with thread cells, and runs out into four short perradial lobes (fig. 4 in the
middle).

The peripheric coronal intestine (" gaster coronaris "), which in most Acraspedae is
divided into from four to sixteen radial pouches or canals, forms a simple wide coronal
sinus (" sinus coronaris," cs) in Tesserantha as in Periphylla. It occupies the whole space
between the septal nodes and the umbrella margin (figs. 2, 3, 5, cs). The broad gastral
openings only may, therefore, be considered as homologous with the four radial pouches
(6p) ; in fact, the four short septal nodes which divide the gastral openings correspond

REPORT ON THE DEEP-SEA MEDUSA. 53

to the four longer septa or iuterradial selvages, which separate the four broad radial
pouches in the Lucernarida? (comp. Pis. XVI., XVII.).

The four reproductive glands (figs. 2, 3, 4, 6, o) in Tesseraniha, as in Tessera and
Depast rum ( = Carduella) form four horseshoe-shaped swellings in the subumbral wall of
the coronal sinus. Their central arch encloses the four septal nodes with its concave
distal margin, whilst its concave proximal margin projects above into the central gastral
cavity, and appears fimbriated with the lowest distal group of the gastral filaments. The
two limbs of the U-shaped reproductive arches run upwards, diverging slightly into the
subumbral wall of the coronal intestine, and their limbs, which are curved outwards,
touch the proximal margin of the coronal muscle. In Tesserantha, as in Tessera, a
thickened gelatinous selvage of the fulcral plate of the subumbrella appears to project
into the hollow space of the coronal sinus, in the entire extent of the reproductive swell-
ing, and the sexual cells appear to be developed from the endodermal epithelium of the
sinus on the axial side of these genital selvages. The spirit specimen (a female) which I
examined did not admit a closer investigation of its finer structure. Slight transverse
folds are visible on the endodermal upper surface of the genitalia (figs. 3, 4, s).

Family, Luceekarid.i;, Johnston, 1847.
LucERXARiDiE, Hajckel, System der Medusen, 1879, p. 379, taf. xxii.

Stauromedusse, wdth lobed or incised umbrella margin, divided by eight concave
arches (four perradial and four iuterradial) into eight hollow- adradial lobes or arms ; a
brush-shaped bunch of hollow knobed tentacles at the end of each arm ; eight principal
tentacles (four perradial and four interradial) either transformed into adhesive marginal
anchors or wanting (having undergone retrograde metamorphosis, or lapsed). Coronal
muscle of the umbrella margin divided into eight isolated marginal muscles. A peduncle
for adhesion at the apex of the umbrella.

Sub-family, Haliclystid^;, Hgeckel (EleutherocaepiD/E, Clark).

Lucernaridse without mesogonial pouches in the subumbral wall of the four radial
pouches.

Lucernaria,1 0. F. Mtiller, 1776.

Lucernaricke without mesogonial pouches in the subumbral wall of the four radial
pouches, and without marginal anchors or marginal papillae (Lucemaridce eleuthero-
carpce inauriculata). Peduncle one chambered with four separate tamiola.

' Lucernaria, like a candle ; from Lucerna, a candle.

.">4 THE VOYAGE OF H.M.S. CHALLENGER.

The genus Lucemaria, the oldest known form among the Stauromedusse, was founded
by 0. F. Muller, more than a hundred years ago, for the large and widely distributed
Acraspeda of the North Atlantic Ocean, which he named Lucemaria quadricomis (Prodom.
Zool. Dan., 1776, p. 227). This notable and well-known form alone can therefore be taken
as the typical species by which to define the character of the genus Lucemaria. Later
authors, such as Keferstein, Leuckart, Taschenberg, and others, have, for the most part,
placed all other later known Lucernaridae in this genus. However, for reasons given in
the System der Medusen (1879, pp. 380, 387), I considered the division of the true
Lucernaridaa into four genera justified, and therefore limit the genus Lucemaria to the
forms without mesogonial pouches and without marginal anchors (EleutherocarjndcB
inauriculata) . The genus Lucemaria is distinguished from the closely allied genus
Haliclystus (which is very minutely described by Clark in his monograph on Halidystus
awricula, 1878) by the absence of marginal anchors or marginal papillae, and from the
two other genera of the family (Halicyathus and Craterolophus) by the absence of the
peculiar " mesogonial pouches or mesenteric pouches," which in the latter penetrate from
the stomach into the subumbral wall of the four perradial gastral pouches. The species
described below (from 3240 feet deep) is the first deep-sea Lucernarid, as all the other
species of this family hitherto described, are littoral, or only found at moderate depths
(from 20 to 50 feet at most). This species is, moreover, distinguished in several ways
and by many peculiarities (especially by the slight development of the eight arms and
the complicated structure of the genitalia) from the four other hitherto known species of
this genus, so that it is perhaps better to constitute it the representative of a special
genus, Lucemosa.

Lucemaria bathyphila, Haackel (Pis. XVI., XVII.).

Lucemaria bathyphila, Hjeckel, 1880, System cler Medusen, p. 640, No. 597.
Lucemosa bathyphila, Hceckel, 1880 (in litteris).

Umbrella bell-shaped, when extended, nearly as broad as high. Peduncle almost rudi-
mentary, conical, one-chambered, hardly one-sixth so long as the cavity of the umbrella,
with four strong, linear, interradial, longitudinal muscles. Eight arms united in pairs.
The four perradial arches of the umbrella margin three times as broad and deep as the
four interradial arches; each arm with 80 to 120 tentacles; eight genitalia, very broad,
separated by a broad interspace from the base of the peduncle as well as from the
end of the arm, each genitalium composed of numerous (above 200) separate sacs,
which again consist of a large number of isolated follicles. Horizontal diameter of the
umbrella, 50-60 mm. ; vertical diameter, 60-80 mm.

Habitat. — The North Atlantic Ocean, between the Faroe Islands and the Shetland
Islands. Lat. 60° 3' N., long. 5° 51' W. Depth, 540 fathoms (John Murray). This species,

REPORT ON THE DEEP-SEA MEDUSAE. 55

the first Lucernarid from the deep-sea, was not taken during the Challenger expedition,
but was part of the spoil of a subsidiary cruise in H.M.S. " Knight Errant," organised by-
Sir Wyville Thomson in the summer of 1880, with the view of verifying some of the
Challenger results. I am obliged to him for giving me an opportunity of including this
species in the list of Challenger Deep-sea Medusae, as in many respects it has a
peculiar interest as a link between the preceding Tesserantha and the following
Periphylla.

The umbrella (PL XVI. figs. 1-8) is of a roundish bell shape, or almost pyriform,
only a little longer than broad, and adhering by a very short peduncle at the aboral pole.
The whole length (or height) (including the peduncle) of the spirit specimen examined
came to 60 mm., the greatest breadth (in the middle of the height) to 50 mm. As,
however, the specimen was strongly contracted, the height in the living animal would
come to at least 70-80 mm., and the breadth to 55-60. This species, as well as the two
closely allied species, Lucernaria quadricornis, and L. pyramidalis belong to the largest
species of the family Lucernaridse ; the latter has a much shorter stem but a smaller cup.

The peduncle (" pedunculus," p>), by which the bell-shaped cup fixes itself to the
bottom of the sea, is rudimentary and slightly developed in Lucernaria bathyphila, as in
all other species of the family. It rather resembles the " apical process or conical process "
of the Tesseridae, from which it is probably derived (System der Medusen, 1879, p. 365,
taf. xxi., xxii.). Its length amounts, at most, to one-sixth of the whole length of the
body, but cannot be sharply defined, as the thicker oral end of the club-shaped peduncle
passes gradually, without distinct boundary, into the cup. The thinner aboral end is
truncated, and has a small roundish disc on the surface of the point of adhesion (fig. 8).
This plate has numerous adhesive cells (" colletocystae ") in its thickened exoderm, it
lies in irregular folds, and is divided by four deep interradial furrows into four perradial
swollen lobes (fig. 8). Each furrow passes a little way into the exumbrella of the
peduncle, so that it also appears four lobed in a transverse section above the disc
(fig. 13). The four interradial, longitudinal furrows of the exumbrella of the peduncle
have four corresponding gastral taeniola in its inner wall (figs. 1, 2, 21, ft); these are
the important longitudinal, gelatinous selvages, already found in Scyphostoma, which
traverse the entire length of the peduncle, and pass immediately below into the four
interradial septa of the gastral pouches (fig. 12, Jos). In the horizontal section these
tseniola appear almost egg-shaped, compressed laterally, and only connected (as by a
peduncle) by a very thin gelatinous plate (fig. 14, ft) with the wall of the umbrella
peduncle, from which they project centripetally inwards. The gastral hollow space of
the peduncle is thus divided into four perradial peduncle grooves (fig. 13, cp) which
communicate by narrower clefts with the central basal stomach (gb), and form a regular
maltese cross seen in transverse section. The peduncle in our species is, however, one-
chambered, as in all species of the genus Lucernaria (in the stricter sense). (System der

56 THE VOYAGE OF H.M.S. CHALLENGER.

Medusen, 1879, p. 389). The four strong tseniola contain a well-developed, longitudinal
muscle (figs. 13, m ; 14, m) ; this is enclosed in a voluminous gelatinous sheath (ft), which
is considerably thicker on the axial side than on the abaxial side, and which internally
forms numerous dendritically branched folds. The muscular plate of the peduncle (m)
is extended on these folds of the gelatinous supporting plate ; it encloses a central axial
cord of exoderm cells (q) lying in the axis of the tseniolum ; there are the " epithelial
muscular cells " of the exumbrella which have immigrated centripetally from its outer
surface into the gelatinous selvage. The delicate figure shown by each tamiolum in its
oval transverse section (fig. 14) consists, from its axis towards its periphery, of the
following layers : — (l) The central cell-cord of the ectodermal epithelial muscular cells
(q) ; (2) the folded muscular plate arising from it (m) ; (3) the fulcral lamella (z) with its
dendritic supporting folds, and the thick gelatinous sheath surrounding it (ft) ; (4) the
endodermal covering of the gastral epithelium (d).

The cup ("calyx"), or the peculiar "umbrella" of our Lucernaria (after removing
the stalk) is almost oval, broadest in the middle, gradually passing into the conical
peduncle above, and slightly contracted below towards the umbrella margin and the
eight arms (figs. 1-3). As in all Stauromedusse, the umbrella consists of two thin walls,
an external convex umbrella and an inner concave suluimbrella. The two walls enclose
the hollow space of the gastrovascular system, pass into each other at the umbrella
margin, and are otherwise only connected with each other by the four interradial septa
(" lines of fusion, or cathammal selvages," h). The two walls consist in section chiefly
of a thin but firm gelatinous plate (fulcral lamella, z) ; its inner side is covered by gastral
endoderm (d), its outer side by dermal exoderm (q). The external convex surface of the
umbrella or the true exumbrella (e) is smooth, without special characteristics, and only
traversed by four slight interradial longitudinal furrows (the distal processes of the
peduncle furrow). The gelatinous substance under the exumbrella is not thick but very
firm, and traversed by numerous elastic fibres which run from the outer to the inner
surface of the gelatinous plate (fig. 13, uf) ; they are also equally numerous in the thin
gelatinous plate of the subumbrella (fig. 18, uf). The ectodermal epithelium of both
the exumbrella and the subumbrella is armed with scattered urticating organs (comp.
my System cler Medusen, 1879, p. 382).

The antrum or umbrella cavity (" necto-calyx ") (h), which is lined by the ectoderm
of the subumbrella (qw), is divided in our species, as in all Lucernaridse, into two parts,
the lower (distal), simple, coronal umbrella cavity, and the upper (proximal), quadrilocular,
funnel umbrella cavity. The coronal umbrella cavity (fig. 5, he, " antrum coronarium ")
is perfectly simple, cylindrical, or almost cubic, and occupies the entire lower half of the
body ; the eight deltoid muscles, and the distal halves of the genitalia he in its sub-
umbrellar wall. The funnel umbrella cavity (" antrum infundibulare," i) is divided from
the coronal umbrella cavity by the oral boundary-line E F (figs. 2, 3), and is composed of

REPOKT ON THE DEEP-SEA MEDUSAE. 57

four interradial pit-like depressions (figs. 6, 7, it) separated from one another by four
perradial vertical folds of the subumbrella (fig. 3, wr). These are Clark's " circumoral
buttresses;" they extend in the form of four free mesenteric lamella? from the four
perradial angles of the oesophagus to the middle of the subumbral wall of the radial
pouches, and are best described as reproductive folds or genital mesenteries (" mesogonics ").
The four depressions lined with the ectoderm of the subumbrella are the funnel cavities
(" infundibula," figs. 2, 6, it). The ctecal ends of these conical or trigonal pyramidal
hollow spaces penetrate from the coronal cavity of the umbrella deep into the central
gastral cavity, and have occasioned many misapprehensions. Clark calls them "circumoral
pouches," Taschenberg " genital pouches," Kling " pyramidal spaces," and Hertwig
" intergenital pouches." As these funnel cavities are only lined by the ectoderm and
have no connection with the gastrovascular system, but belong much more to the
system of the subumbral umbrella cavity, they cannot be termed " pouches " but merely
" cavities." They recur in the same way in many other Acraspedse as " subgenital
cavities." In our Lucernaria they penetrate so deep into the central gastral cavity as
to divide its oral half into four perradial peripheric niches, or " central chambers."
The conical funnel cavities between the latter are separated from them by the gastral
filaments, and pass directly above into the solid tseniola (fig. 21, ft).

Tbe muscular plate of the subumbrella lies immediately under the ectodermal epithe-
lium, from which it is secreted, and consists of a marginal octomeral coronal muscle and
of eight separate radial muscles. The coronal muscle ("musculus coronarius"), or circular
muscle of the umbrella margin, is homologous with the simple marginal circular muscle of
the Tesseridse and with the large octomeral coronal muscle of the Pericolpidse, which in
the Periphyllidas is divided into sixteen muscular area?. The coronal muscle in the eight-
armed Lucernaridae, as in the closely-allied eight-lobed Pericolpicke, consists of eight
separate arese, the eight "marginal muscles," of which four longer (figs. 2, 3, 12, mm)
lie in perradial octants, four shorter (mm" ) in four interradial octants ; as, however, the
eight arms (or marginal lobes) are adradrial, each coronal muscular area (or each
marginal muscle) applies to the two halves of each two adjacent arms turned to
each other. It extends on their external or abaxial side, and that of the tentacles
running out from them. The separate bundle of muscles, which here pass into the
tentacles, therefore extend them and make them arch outwards (" extensores"). If, on
the other hand, all the eight marginal muscles contract simultaneously, they narrow
the umbrella opening like the simple circular muscle of the Tesseridae. The coronal
muscle is, moreover, in all Lucernaridae much narrower than in the Pericolpidae, and
has the form not of a broad band, but of a thick cord. In our species this cord
shows six to eight deep parallel furrows, divided from each other by the same number of
circular folds (fig 20 in radial transverse section). The height of these folds increases
from above downwards (from the proximal to the distal margin of the marginal muscle).

(ZOOL. CHALL. EXF.— PART XII. — 1881.) M 8

58 THE VOYAGE OF H.M.S. CHALLENGER.

Each fold is formed by an elevation of the supporting lamella (2), which again forms
secondary folds, and therefore appears dendritically branched in transverse section.
The circular fibres of the muscular plate (m) cover this system of folds connectedly ; and
are covered in their turn by the ectodermal epithelial cells of the subumbrella (qw) from
which they are secreted. The eight longitudinal deltoid muscles (figs. 2, 3, 4, 12, md)
work antagonistically to the eight circular marginal muscles. In the deltoid muscles of
our species the four perradial (md') are very weak but very broad, whdst on the other
hand the four interradial (md") are much narrower, but proportionately more strongly
developed. These appear to be the direct processes of the strong tseniola muscles,
they run along the entire length of the septa of the pouches (fig. 12, tr.s), and split up
below at the distal end of the latter, into two strong limbs (fig. 12, md'") each of which
bears a bunch of tentacles.

The umbrella margin has eight shallow concave depressions or " marginal sinuses,"
between which, as in all Lucernaridse and Pericolpidse, lie eight adradial marginal lobes.
These eight adradial hollow marginal appendages, which have hitherto been generally
termed "arms" in the Lucernaridse and erroneously considered a special peculiarity of this
family, are, in fact, from their situation, structure, and signification, merely the eight
adradial marginal lobes of the closely allied Pericolpidae and as such homologous to the
eight sense lobes (or "eye lobes") of the Peiiphyllidae. The essential difference from
the Pericolpidas, which is strikingly displayed by the Lucernaridse, is that each of the
marginal lobes or " arms" bears a brush-shaped bunch of numerous small, hollow
knobbed tentacles at their points. Morphologically considered, these tentacles belong to
the category of accessory or secondary tentacles, and are merely long-stalked urticating
knobs. On the other hand the four principal tentacles of Tessera (four primary perradial
and four secondary interradial) have disappeared in the genera Lucernaria and Cratero-
loplms, whilst in Haliclystus and Halicyathus they are transformed into adhesive
" marginal anchors." In our species the eight arms are very small, and less developed
than in most other Lucernaridas ; they project only slightly from the umbrella margin
as broad triangular points and are placed together in pairs, so that the four perradial
sinuses of the umbrella margin are three times as large as the four interradial (figs. 1-4).
Each short arm or marginal lobe bears a bunch of from 80-120 tentacles.

All the tentacles (PI. XXII. figs. 15, 16) are completely fused together at their
basal halves, so that only their distal halves are free and movable (fig. 15). They
are cylindrical, 2-3 mm. long, when contracted (probably twice as long when extended),
and nearly \ mm. thick. As in all true Lucernaridse, they are hollow, thick-walled little
tubes, whose csecal and somewhat thinner distal end bears a thicker urticating knob.
This stalked urticating knob in our species is developed into a strong sucking-cup, with a
depressed sucking-pit in the middle (fig. 16, x). The hollow cylindrical epithelium of the
ectoderm (q) is four to six times as high in the sucking-cup as on the tentacle stalk, and

REPORT ON THE DEEP-SEA MEDUSAE. 59

has a peculiar construction, which however could not be more minutely investigated as
the tentacles were not sufficiently well preserved. The ectodermal cells in the central
sucking-pit are much flatter, and without nematocysts (fig. 16, x). The muscular plate,
consisting of strong longitudinal fibres (m), lies immediately under the ectoderm (q).
Then comes a thick gelatinous supporting plate, which acts as elastic extensor against
the pull of the longitudinal muscular fibres, and re-extends the contracted tentacles,
shortened by the latter. In the sucking-cup the gelatinous fulcral lamella forms a
peculiar thick cap (fig. 163'), which encloses the csecal end of the tentacle canal and is
sharply divided by a distinct boundary line (z") from the thinner gelatinous plate of the
peduncle (z"). The endoderm (d) forming the epithelium of the tentacle canal (ct),
consists of high, dark-brown pigmented cylindrical' cells, having numerous unicellular
glands distributed among them. The csecal distal end of the tentacle canal shows a
very peculiar condition unknown to me in any other Lucernarid. The end of the
canal is closed by a conical wedge, which completely fills the distal end of the lumen of
the tube and is enclosed by endoderm. This axial wedge of the sucking-cup (fig. 16, y)
dyes a much more intense dark red with carmine than any other part of the tentacle.
It contrasts sharply with the yellow-brown endodermal cells enclosing it, and seems
composed of roundish corpuscles, thickly pressed together, which refract light strongly
and look like nematocysts. This axial wedge of the sucking-cup perhaps serves as a firm
support during its adhesion.

The formation of the gastrovascular system (PL XVI. figs. 2-7; PI. XVII. figs. 13-16)
does not differ essentially in our Lucernaria from that known in other species of this
genus ; it lies between the simpler formation of the Tessericlse (PI. XV.) and the more
developed formation of the Periphyllidae (Pis. XVIII. -XXII.). As in the Tesseridse it is
divided into a central princij)al intestine (" gaster principalis "), and a peripheric coronal
intestine (" gaster coronalis "), communicating by four perradial gastral openings (go).
The principal intestine consists of three parts, viz., the aboral basal stomach in the
peduncle, the central stomach, and the freely projecting buccal stomach or oesophagus.
The central stomach is separated from the basal stomach by the pyloric opening (" pylo-
rus"), and from the buccal stomach by the palatine opening ("palatum"). The basal
stomach (" gaster basalis," (76) is the peduncle canal already mentioned; it passes through
the entire length of the peduncle, and ends cascally in its oral basis, whilst it opens at
the oral peduncle end by the pylorus (gy) into the central stomach. The basal stomach
originally presented a simple cylindrical or Cjuadrangularly prismatic hollow space, corre-
sponding to the " apical canal" of the Tesseridse. As the four interradial tseniola (ft)
project from the wall of the peduncle into the stomach, they divide its periphery into the
four perradial peduncle grooves or half canals already described (fig. 13, cp). In this
way the basal stomach acquires in transverse section the characteristic regular cross
shown in fig. 13, Plate XVII. The central stomach (" gaste centralis ") has usually a coni-

60 THE VOYAGE OF H.M.S. CHALLENGER.

cal or quadrangularly pyramidal shape, and opens with the truncated aboral end into the
peduncle stomach by the pyloric opening (" pylorus," gy) and with the cpuadratic, strongly
constricted oral basis into the buccal stomach by the palatine opening (" palatum," gp).
Four perradial clefts, the gastral openings (" ostia gastralia," figs. 2, 3,gp ; fig. 21, go) lead
from the central stomach into the four radial pouches. As the four, conical interradial
funnel cavities (ii) already described project arch-like between the four gastral openings
into the central stomach, the formation of the latter becomes rather complicated. Both
margins of the cleft-shaped gastral openings are edged nearly their whole length by a row
of fine gastral filaments (fig. 21,/) ; these are only wanting on the lower (oral) fourth of
the gastral openings where their margin forms the cartilaginous-like thickened groove of
the palate (fig. 21, gs). The rows of filaments or phacelli run above to the point of the
funnel cavities, but do not pass on to the solid tseniola. In comparison with other
Lucernaridse, and with the considerable size of our species, its filaments appear slightly
developed, very fine and rather short ; they are limited here to the lateral margin of the
gastral openings, whilst in other species they often extend distally far on to the lateral
margins of the genitalia, or proximally on the basal tseniola. Like the filaments, the oeso-
phagus or " buccal stomach " seems only slightly developed in Lucemaria bathyphila
(proboscis, figs. 2-4, ga). It forms a low, fleshy, membraneous border, quadrate in outline,
which only projects sbghtly from the palatine opening into the umbrella cavity. The
thickened glandular margins of the oral opening are only slightly frilled (fig. 9). The
four perradial angles of the oesophagus pass at the palatine opening, into the four sub-
urnbral mesogonial folds (ivz).

The peripheric coronal intestine (" gaster coronaris "), which only communicates by the
four perradial gastral openings with the central stomach, in Lucemaria bathyphila is
formed (owing to the slight development of the eight arms) almost exclusively by the
four voluminous radial pouches (" bursse radiales," bp) which Clark termed " quadrant
chambers," Keferstein " broad pouch-shaped radial vessels," Taschenberg " radial canals,"
KJing "radial chambers," and Hertwig "radial chambers or radial pouches." These
present four flat pouch-like hollow spaces extending between the umbrella and subum-
brella to the umbrella margin. They are only divided by four interradial " septal
selvages," or "lines of fusion," linear septa in wThich the umbrella is fused with the
subumbrella (" septa cathammalia," ks). As this fusion does not, however, extend as far
as the umbrella margin, the four pouches communicate there below the distal end of the
septa, by four interradial circular openings, so that a circular communication, a sort of
"circular canal," is formed on the umbrella margin (fig. 12, cc). The proximal half of
the four radial pouches opens by the gastral openings into the central stomach ; whilst
eight lobe pouches or " arm pouches " (" bursas lobares ") run from its distal margin into
the eight arms. The end of each lobe pouch again sends out a tentacle canal into each
tentacle (figs. 15-16, ct). As the eight arms or marginal lobes in our species project onlv

REPOKT ON THE DEEP-SEA MEDUSAE. 61

slightly above the umbrella margin, their lobe pouches (bl) never reach an independent
development. The development is consequently proportion-ably greater of the four
radial pouches, whose length is nearly two-thirds that of the whole length of the umbrella,
and only partially occupied by the conspicuous genitalia, lying in their subumbral wall
(comp. figs. 2-7, bp).

The genitalia (PI. XVI. figs. 2-7, s, figs. 10, 11; PI. XVII. figs. 17-19, 21). The
specimen examined was a mature female, and showed most distinctly that the ova in the
Lucernaridse (as in all Stauromedusse) are developed from the endoderm of and in the sub-
umbral wall of the radial pouches, then fall into their cavity, reach the central stomach
through the gastral openings, and are finally expelled through the mouth ; all parts of the
gastrovascular system of the uninjured Medusa contained when opened numerous, loose,
ripe ova. The ovaries (figs. 2, 3, sf) form eight broad plates occupying the greater part
of the subumbral wall of the four perradial gastral pouches, and are distributed in pairs in
such a manner that the two genitalia separated by an interradial septum, form a connected
pair. The two ovaries lying in one and the same radial points, therefore form two
different pairs (comp. my System der Medusen, 1879, p. 386). The interradial interval
between each two reproductive leaves is considerably smaller than the perradial interval ;
their distance from the distal margin of the four radial pouches is also much less than
from the proximal margin (comp. figs. 2, 3, sf). Their outline is semi-oval or almost
lanceolate, and broadened in the distal third.

The structure of the ovaries in Lucernaria bathi/phila is very peculiar, and more
complicated than in all other Stauromedusse hitherto known. Even with the naked eye
the upper surface of the eight reproductive glands appears granular as if paved, and a
slight magnifying power (fig. 21, sk) shows that each genitalium is composed of a large
number (nearly 200-250) of entirely separate sacs. These have an irregular roundish or
polygonal outline, averaging 1 mm. in diameter (the smallest rather under ^ mm., the
largest rather above 1-^ mm.). Whilst in all other Lucernaridse hitherto known these
reproductive sacs (" sacculi genitalis," sk) represent simple glands with a single hollow
space and excretory passage, in our deep-sea species they are lobed glands composed of
several separate lobes or follicles, each having its own cavity and its own excretory
passage. Each separate sac (fig. 10, seen from the surface, fig. 18, in perpendicular longi-
tudinal section) is therefore usually composed of from thirty to fifty follicles (sb). Each
separate follicle (fig. 11 from the surface, fig. 19 in longitudinal section) contains a
"sinulus" (sc) or secondary cavity, which opens by a "ductulus" or secondary excretory
passage (si) into the "sinus genitalis," or principal cavity of the sac (fig. 18, sc) ; the last
opening by its ductus or principal excretory passage (si) into the radial pouch (sa). The
ova (fig. 19, so), which are developed from the endodermal epithelium of the follicle,
pass first from its sinulus (so) into its ductulus (si), thence into the sinus of the sacculus
(fig. 18, sc), and from the sacculus by the ductus (si) into the radial pouch. The ovary of

62 THE VOYAGE OF H.M.S. CHALLENGER.

Lucernaria bathyphila, therefore, furnishes the first example in the order of Stauro-
medusoe of a complicated reproductive gland with lobed sacs and branched hollow spaces ;
and this, and other peculiarities already mentioned of our deep-sea species, justify its
being raised to the type of a separate species, Lticernosa. Both the separate sacs
and the follicles of which they are composed are enclosed in a thin structureless " mem-
brane propria," a direct process of the gelatinous fulcral plate of the subumbrella. The
sacs (figs. 17, 18, sk) project freely from the subumbral wall of the radial pouches, on
whose endodermal surface they are placed, into the hollow space of the pouches ; their free
abaxial surface is covered by the ciliated endodermal pouch epithelium, whilst their
fixed axial surface is divided from the ectodermal pouch epithelium (qiv) by the thick
gelatinous plate of the subumbrella (figs. 17, 18, tig). There is, therefore, no doubt that
the ova are developed from the endodermal cells of the sacs, which has no connection
with the subumbral ectoderm.

Order VI. PEROMEDUS^, Haeckel, 1877.

Acraspedae with four interradial rhopalia, containing an auditory club with an endo-
dermal otolite sac and one or more eyes, four perradial tentacles or twelve tentacles (four
perradial and eight adradial), eight adradial or sixteen subradial marginal lobes. Stomach
surrounded by a subumbral coronal sinus, whose division into four radial gastral pouches is
only indicated by four small interradial septal nodes ; eight or sixteen coronal pouches on
the distal margin of the coronal sinus ; two lateral lobe pouches from each coronal pouch,
and in the middle between the lobe pouches, a pouch for the tentacle or the rhopalium.
Genitalia, eight adradial horseshoe-shaped swellings which lie in the subumbral wall of
the coronal sinus, are developed from its endoderm and partly project into its cavity.

Family, Periphyllid^e, Hseckel, 1877.

PERiPHYLLiDiE, Hfecke], System der Medusen, 1879, p. 415, plate xxiv.

Peromedusas with twelve tentacles (four perradial and eight adradial), with four inter-
radial rhopalia and sixteen subradial marginal lobes (eight tentacular and eight ocular).
Exumbrella with sixteen pedalia, and a coronal muscle with sixteen coronal areae (four
perradial, four interradial and eight adradial), a coronal pouch between each pedalium
and each coronal area. Marginal festoon canal formed of thirty- two lobe pouches.

Sub-family, Pepjphemid^e, Haeckel, 1880.

Periphyllidaa whose four interradial funnel cavities are not limited to the central
stomach but also traverse the basal stomach, wholly or partially.

REPORT ON THE DEEP-SEA MEDUSvE. 63

Periphylla* Steenstrup, 1837.

Periphyllidas with four perradial buccal pouches of the oesophagus and four perradial
completely separated niches of the basal stomach. Between these niches, the four
subumbral funnel cavities (or the four hollow interradial tasniola of the basal stomach)
form hollow cones, which are beset along their whole length by two rows of gastral
filaments and touch each other above in the point of the cone.

The genus Periphylla, as well as the following closely allied genus Periphema, was
only presented in the Challenger collection by a single specimen. However, its large
size and its excellent state of preservation enabled me to examine it more minutely and
thoroughly than I had ever been able to examine any other Peromedusa. So that the
following description of Periphylla mirabilis with the six plates (XVIII.-XXIII.) may
be accepted as a firm foundation for the anatomical knowledge of the whole order of
Peromedusse. This conspicuous and remarkable group of Acraspeda was, till lately,
almost unknown. On the one hand, it keeps in many ways the primitive formation of the
Stauromedusse, and is more closely connected both with the Tesseridae and the Lucer-
naridse than the two orders of Cubomedusse and Discomedusse, especially with regard to
the remarkable formation of the central gastrovascular system. On the other hand, it
is raised so far above the other three orders of Acraspedae by the peculiar complication
of its anatomical structure, and specially by complicated formation of the pouches, that
in many respects it may be called the most highly developed of all Medusae. At any
rate, we must consider them as an independent principal group, as a special " order " of
Acraspedse, which have no direct connection with the Cubomedusae and Discomedusae,
but must be rather regarded as a peculiarly developed branch of the Stauromedusse. All
that was known of the wonderful Peromedusa? up to the year 1879 was limited to the
imperfect description of three different species of the genus Periphylla. But two of
these figures showed only the empty umbrella of the dead animal without any internal
organs — Charybdea periphylla, Peron and Lesueur (1809) ; and Charybdea bicolor,
Quoy and Gaimard (1833). The description of the third species, Dodecabostrycha
dubia, Brandt (1838), is partly good, partly very erroneous and incomplete, and
remained to be completely unintelligible. Detailed examination of several well-
preserved specimens of the stately Periphylla hyacinthina and some other smaller
species made by me on the genera Pericolpa, Pericrypta, and Peripalma first enabled
me, in 1879, to describe more minutely the hitherto unknown organisation of the Pero-
medusse, and to place them as an independent order of the class (in the System
der Medusen, pp. 396-422, Pis. XXIII., XXIV.). The anatomical description given
there will, however, be enlarged and completed in many points by the following more

1 ritp;^i)xA«=set round with leaves.

64 THE VOYAGE OF H.M.S. CHALLENGE!;.

detailed anatomy of Periphylla mirabilis. This species, as well as the following species,
Pervphema regina belong to the family of the PeriphyUidse, these large and highly
organised Peromedusse which have twelve tentacles, sixteen marginal lobes, and thirty-
two lobe pouches. Contrasted with these stand the older and more simply constructed
Pericolpitis;, which are more closely connected with the Lucernaridas, and have only four
perradial tentacles, besides eight marginal lobes, and sixteen lobe pouches. Each perradial
tentacle of the Pericolpidse is represented in the PeriphyUidse by three tentacles, and two
marginal lobes inserted between them. All Perom'edusfe have invariably only four
interradial rhopalia, and this alone distinguishes them from all other Medusae. Our
Periphylla mirabilis is distinguished from all other species of the genus by the eight
strong barbous filaments of the margin of the mouth, and may therefore be regarded
as the representative of a distinct genus — Periphenga mirabilis (irepi^iyya, radiating).

Periphylla mirabilis, Hseckel (Pis. XVIII.-XXIIL).

Periphenga mirabilis, Hreckel, 1879, System der Medusen, p. 422, Xo. 424.

Umbrella conical, about one-fourth higher than broad. Pedal zone of the exunibrella
somewhat higher than the lobe zone, both together nearly three-fourths as high as the
cone zone. Marginal lobes oval, pointed ; their distal wings triangular, half as high as
their proximal gelatinous swelling. The eight tentacle lobes projecting less than the
eight rhopalia lobes on the umbrella margin. Tentacles twice as long as the height
of the umbrella, one-third as broad as the marginal lobes at their basis. (Esophagus
cubical, one-third as high as the umbrella, reaching only to the coronal muscle, with
eight adradial, long, feathered, barbous filaments at the margin of the mouth. Hori-
zontal diameter, 120 mm. ; vertical diameter, 160 mm.

Habitat. — South Pacific Ocean, near the east coast of New Zealand. Lat. 40° 28' S.,
long. 177° 43' E. Station 168. The single specimen captured, a mature male, was
taken July 8, 1874, at a depth of 1100 fathoms. It was admirably preserved in spirit, was
quite perfect ; and was, on the whole, of a pale violet colour. The inner or endodermal
surface of the gelatinous umbrella was overlaid with dark, violet-brown pigment, which
was easily rubbed off, and consisted of small roundish granules in the endodermal cells of
the abaxial wall of the gastral space. The tentacles appeared coloured darker violet, the
genitalia reddish-yellow. After lying some years in spirit, the colours become fainter.

The umbrella (PI. XVIII. fig. 1 ; PL XIX. fig. 6 ; PI. XX. fig. 8 ; PL XXI. figs.
12-20) of Periphylla mirabilis, as of most other Peromedusse, is high-arched, conical,
pointed above, and widened like a funnel below, or almost helmet-shaped. The height (or
vertical diameter) of the umbrella — including the marginal lobes, excluding the tentacles
— amounted in the uninjured specimen to 16 centimetres; therefore, one-fourth more
than the breadth at the opening of the umbrella, 12 centimetres being the largest
horizontal diameter. Almost in the middle of its height, — 8^ cm. from the umbrella

REPORT ON THE DEEP-SEA MEDUSAE.

65

cone, 1\ cm. from the umbrella margin, the umbrella is deeply constricted by a
horizontal circular furrow, the coronal furrow (" fossa coronaris," ec). The umbrella cone
or "cone zone" ("zona conaris") is a perfectly simple smooth cone, pointed above and
widening symmetrically below ; its firm, gelatinous substance is of nearly equal thickness
throughout, about 8 mm. ; in the coronal furrow (ec) its thickness suddenly diminishes
to 2-3 mm. (fig. 35). The external surface of the cone is flat throughout without
any ornament. The corona of the umbrella, on the contrary, is subdivided by a lower
horizontal circular furrow into two zones, the upper pedal zone and the lower lobe zone.

The pedal zone (" zona pedalis ") is divided by sixteen deep longitudinal furrows into
the same number of projecting wedge-like gelatinous socles, the pedalia. Of these the four
interradial are considerably smaller (25 mm. high, 12 mm. broad above, and only 8 mm.
below), and bear below the four sense clubs and their ocular lobes ("pedalia ocularia").
The twelve remaining gelatinous socles (" pedalia tentacularia ") bear the tentacles and
their lobes below, and are much larger (35 mm. high, 13 mm. broad above, 17 mm.
below). Between each two interradial ocular pedalia (fig. 19, ni) there are three larger
tentacular pedalia, the central one («/>) lying perradially, the two lateral (ua) adradially,
(comp. figs. 18 and 19). The firm gelatinous substance of the umbrella is 10-12 mm. in
thickness in the upper part of the pedalia, whilst in their lower part it is diminished to
3-5 mm. Each pedal is limited on the upper convex margin by a crescentic area (" areola
semilunaris") (fig. 34, xs), which is divided by 8-10 small shallow longitudinal furrows
into the same number of smaller swellings (" gyruli") ; they end above in the bottom of
the large coronal furrow, and contain pointed processes of the exumbral zonal muscle
(fig. 34).

The exumbrella of the corona of the umbrella in our Periphylla, as probably in all Pero-
medusse, is distinguished, not only by the longitudinal and transverse furrows already men-
tioned, which penetrate more or less deeply and are connected with important internal
anatomical and ontogenetic conditions of organisation, but also by special exumbral
muscles. In the deep coronal furrow between the umbrella cone and umbrella corona
(" fossa coronaris," ec) there is a ring-shaped external zonal muscle (" musculus zonaris,"
mz), with sixteen exumbral points projecting from its distal margin ; these are triangular,
having the point directed downwards and correspond to the radii of the marginal lobes ;
they consequently lie subradially, and their distal end passes below in the furrow,
between the two pedalia in whose prolongation the lobe clasp (fig. 34, mz) is placed. Both
the zonal muscle itself and the muscular points proceeding from it are composed of strong
annular fibres.

The true umbrella margin (in the wide sense) (" margo umbralis," um ; PI. XVIII.
fig. 1 ; PL XIX. fig. 6 ; PI. XX. fig. 8 ; PI. XXII. fig. 22) is formed by the lobe zone of
the umbrella corona already mentioned, and consists of the following important organs : —
four interradial sense clubs, twelve tentacles (four perradial and eight adradial), and

(ZOOL. CHALL. EXP. PAET XII. 1881.) M 9

66 THE VOYAGE OF H.M.S. CHALLENGER.

sixteen subradial marginal lobes, inserted between the former and the latter. These organs
in all Peromedusee, show very peculiar and complicated conditions of structure, which
thoroughly distinguish them from both the Stauromedusse and the Cubomedusse. Notwith-
standing, these structures are phylogenetically derived from those of the Stauromedusse
and then from those of the Pericolpidse (comp. my System, 1879, taf. xxiii.). The
four perradial tentacles and their four interradial sense clubs have arisen from the eight
principal tentacles of the Tesseridaa, and are therefore homologous with the "marginal
anchors" of the Lucernaridse ; on the other hand the eight adradial tentacles with the
alternating marginal lobes of the Pericolpidse, are homologous with the hollow arms of the
Lucernariche. The umbrella margin of the Periphyllidse has plainly arisen in this way
from that of the Pericolpidse, as instead of each perradial tentacle, there are three
tentacles with two additional marginal lobes between them. The number of tentacles
rises in this way from four to twelve, and the number of the marginal lobes from eight
to sixteen. The original number of the four sense clubs remains invariable in all Pero-
medusae, and is typical of the whole order.

The sixteen marginal lobes ("lobi marginales," PL XVIII. fig. 1 ; PI. XXII. fig. 22 ;
PI. XXIII. figs. 29-32) of the corona of the umbrella are, on the whole, egg-shaped, lie
subradially in the meridian plane of the fourth order, and are divided into four pair of
ocular lobes (lo), and four pair of alternating tentacular lobes (It). The two ocular lobes
(" lobi oculares ") of each umbrella quadrant lie exradially, as the interradial eye is set
between them. The marginal incision between them up to the eye is 17 mm. deep, half
as deep as the incision between each ocular lobe and the neighbouring tentacular lobe.
The two ocular lobes of each pair therefore compose a broadly oval, ocular principal lobe,
whose free margin is divided into two secondary lobes (fig. 1). Each ocular secondary
lobe is divided by a deep longitudinal furrow into two halves, an adocular and an exocular.
The exocular or external half, which touches the adradial tentacle, is wing-shaped and
thinned, and runs into a delicate membranous selvage in the margin of the lobe
(" patagium," Ip). The inner or adocular half, which touches the interradial eye, is
strongly thickened, so that in conjunction with that of the adjacent secondary lobe, it
forms a thick swelling, projecting convexly outwards, in direct prolongation of the ocular
pedalium (ur). Each ocular principal lobe therefore appears like a broad oval leaf, which
has a midrib 1 cm. broad, side-wings 3 cm. bi'oad, and is 4 cm. long in all (in the
middle line). The two tentacular lobes ("lobi tentaculares ") of each quadrant of the
umbrella lie coradially, as they enclose the perradial tentacle beween them and are
divided externally from the ocular lobes by the adradial tentacle. Each of the two
tentacle lobes represents a longish oval leaf 4 cm. long and 2 cm. broad, divided by a deep
subradial longitudinal furrow into two lateral swellings ; this furrow forms the direct pro-
cess of the coradial furrow, which separates the adradial pedalia of the furrow zone from
the perradial. Each of the two swellings of each tentacular lobe is almost linear. A very

REPORT ON THE DEEP-SEA MEDUSAE. 67

thin, folded, wing-shaped membranous selvage (" patagium," lj)), 5-8 mm. broad, also
runs round the margin of this lobe ; it is broadest at the point and narrowest at the base
(fig. 1, 22, &c).

The fused clasp of the marginal lobes (" loboporpa, cathamma lobare," hi; PI.
XXII. fig. 22 ; PL XXIII. fig. 29). The exumbral longitudinal furrow of each lobe,
which divides its two gelatinous swellings and passes above into the" sulcus interpedalis,"
has a corresponding fused clasp (hi) in the interior of the lobe. This is a rectilinear
gelatinous selvage of cartilaginous hardness, firmness, and elasticity. It springs with a
broad basis (fig. 29, hi) from the proximal margin of the coronal muscle, and reaches the
border of the middle and distal third of the lobe, where it becomes thicker and ends (fig.
22, hi"). Its peculiar structure is shown in Plate XXV. fig. 10, under a higher magnifying
power. The clasp is formed by the umbral or abaxial endodermal epithelium (fig. 10, dw)
and the subumbral or axial endodermal epithelium (fig. 10, dw2) becoming fused into
sixteen subradial straight lines in the peripheric part of the circular sinus, by which the
latter is divided in the region of the coronal muscle into sixteen coronal pouches (be).
The fused clasp of the marginal lobes completely divides the two adjacent coronal
pouches, but only partially divides the lobe pouches proceeding from them (PI. XXV.
fig. 10, bl), which are bent into each other like a horse-shoe at the thickened distal end of
the clasp (bu, figs. 22, 29). Both the thicker gelatinous plate of the umbrella
(PL XXV. fig. 10, ng) and the thinner gelatinous plate of the subumbrella (zw) undergo
considerable induration and peculiar histological change in the region of the fusion of the
two layers of endodermal epithelium. The soft gelatinous substance becomes a firm
fibrous cartilage with numerous roundish cells separated by a fibrillar intersubstance.
The fibrous cords of the intercellular substance cross each other in all directions, as they
do in the analogous cathamma of the nodes of the septa (PL XXV. fig. 4, hi).

The sixteen tentacles (PL XXVIII. fig. 1 ; PL XIX. figs. 6, 7 ; PL XXI. fig. 21 ; PL
XXII. fig. 22) are strong, hollow cylindrical tubes, which gradually decrease conically
towards the distal point. All the twelve tentacles (four perradial, eight adradial) are of
equal size. They are from 30-40 cm. long, consequently twice the height of the
umbrella ; they are possibly 50-60 cm. long in the living animal. The tentacles are
8 mm. thick at their enlarged conical bases ; 3 cm. below the insertion they are 5 mm.
thick ; 5 cm. below only 3 mm., and so gradually decrease towards the point which runs
out almost to a thread. The smooth upper surface of the cylindrical tube appears
repeatedly constricted by numerous annular folds, which are only interrupted by the
longitudinal muscles (figs. 1, 6, 7). The spacious cavity of the tentacles is enclosed by a
thin but very firm leather-like wall.

The tentacle wall is formed of four layers (PL XXI. fig. 21):— (1) the endodermal
epithelium (dr) of the canal (ct) ; (2) the supporting plate (z) ; (3) the muscular plate (m) ;
(4) the ectodermal epithelium of the outer surface (q). The finer structure of the wall is

C,s THE VOYAGE OF H.M.S. CHALLENGER.

very peculiar, but could not be diagnosed satisfactorily from the only spirit-specimen
examined. The ectodermal epithelium (q) contains numerous nematocysts. The muscular
plate (m) appears thinned away on both lateral surfaces of the tentacles, but on the other
hand thickened so remarkably on the inner and outer side that it projects in the form of
two strong band-shaped longitudinal muscles. The external or axial longitudinal muscle
springs from the pedalia, and usually occupies only the proximal third or fourth of the
length of the tentacle. The inner or axial longitudinal muscle runs through the entire
length of the tentacle, and is split up above into two conical root muscles (ink, figs. 22, 29).
These invaginate the distal margin of the corresponding coronal pouch, divide it into an
external velar pouch and an internal avelar pouch, diverge into the " funnel of the
tentacle " (it), formed in this way between the velar and avelar pouches and run as the
proximal margin of the coronal muscle, where they are inserted (comp. below). If the
internal longitudinal muscle is strongly contracted, the tentacle appears rolled up spirally
and laid in deep transverse folds (d). A structureless thin, but very firm, supporting
plate ("lamina fulcralis," fig. 21, z) lies under the muscular plate (m) and under the
supporting plate, the enclodermal epithelium of the tentacle canal (d). The nature of the
latter is very remarkable ; it consists of large vesicular cells, and rises in the shape of a
thick spongy cord (fig. 21, d") on the abaxial side of the canal wall. This cord consists
of an accumulation of very large vesicular cells, and fills like marrow nearly the half of
the lumen of the tube (ct). It would recpiire to be more minutely investigated in living
and well-preserved animals. So would another most peculiar arrangement of the ten-
tentacles ; a strong, double- valved aperture lying inside the base of the tentacle immedi-
ately at the point where the tentacle roots diverge (PL XXII. fig. 22, yk). The elastic
fulcral lamella is swollen there into a thick gelatinous plate containing cells, and forms two
horizontal vent-valves lying above one another, by means of which the cavity of the
tentacle can be completely closed. Even by strong pressure from within the tentacle
cavitv it was impossible to overcome the antagonism of the double valve. The cavity of
the vent-hole (fig. 22, ex) between the distal (yk") and the proximal valve (y¥) is nearly
as hioh as broad.

The marginal sense clubs (" rhopalia") of Periphylla (PI. XVIII. figs. 1-5 ; PL XXII.
fig. 22, or; PL XXIII. 31, 32, or) have been already described by me in Periphylla hyacin-
th i 'nam my System der Medusen (1879, taf. xxiii. figs. 9-12). They appear to have
essentially the same formation in Periphylla m irabilis, and represent very composite organs
of sense connected among the forms hitherto known, with the rhopalia of the Cubornedusaa
on the one side and with those of the Nausithoidae on the other. As in all Peromedusa?
there arc only four interradial rhopalia, which lie in the radii of the septal nodes and
the tseniola. They were, unfortunately, very badly preserved in the spirit-specimen
examined ; a complete and correct insight into their very complicated minute structure
could only have been obtained by examination and special preparation of fresh rhopalia.

REPORT ON THE DEEP-SEA MEDUSAE. 69

With the naked eye they can he recognised as white granules in the incision between the
two ocular marginal lobes. Each rhopalium consists of a conical basal part, the sense
knob, of a large sense vesicle on the axial side of the knob, and of a sense fold or
protective scale which is placed at the distal end of the sense knob and surrounds the
auditory club as well as the eye (comp. PL XVIII. fig. 2, seen from the inside, axial side ;
fig. 3, seen from outside, abaxial side ; fig. 4, seen in profile ; fig. 5, seen half from the
inside, half in profile). The sense knob corresponds to the basal part of the greatly
shortened and thickened tentacle, from which the whole sense club is phylogenetieally
derived. It projects between the bases of its two constituent sense lobes, is usually
conical in shape, and bears the large spheroidal or oval sense vesicle (" ampulla rhopalaris,"
oa), a caecal arching outwards of the sense pouch (bo) on its inner or axial side. Just
under the ampulla the sense knob is constricted like a neck and surrounded by the large,
darkly pigmented sense collar (op). The latter forms an ectodermal swelling, with a thick
accumulation of brown or dark pigment and has cmite the shape of a high coat collar,
which closes round the neck of the sense knob on the abaxial (external or dorsal) side,
whilst it falls obliquely on the axial (inner or ventral) side, and passes into two lateral
symmetrically-placed lapels (figs. 2, 3, op). The two lapels of the sense collar are divided
from one another by a deep, broad furrow, which is only bridged over below at the distal
margin of the collar by a narrow cross bar of pigment. An unpaired axial eye with lens
(oc') and pigment cup appears to lie in the depth of the furrow between the end arms
and the pear-shaped auditory club (ok), the distal end of the acoustic tentacle rises on a
thin stalk immediately below it. The auditory club is white, and consists of an axis of
endoderm cells, the last of which forms a large spheroidal otolite sac, closely filled with
numerous small prismatic crystals (ol). The ectodermal covering of the auditory club
bears auditory hairs which project freely into the niche of the auditory scale (os). The
latter forms a protective scale, oval or triangular in shape, arched convexly outwards,
concavely inwards, so that it surrounds the auditory club as a protection from the
abaxial (external) and distal (lower) side. Two eyes containing a jalaneonvex or
biconvex lens in the midst of a cup of brown or black pigment (?) appear placed inside the
niche of the scale (on) on the abaxial side of the auditory club (between the otolite sac
and the sense collar). All these conditions could, unfortunately, only be indistinctly and
incompletely recognised in the poorly-preserved spirit-specimen, so that it was only by
aid of comparison with the sense clubs of some other Periphylla that I was able to draw
out figures 2-5 reproduced in Plate XVIII. , which can only claim to be approximately
or even remotely correct. It may, however, be safely asserted that the sense clubs
of Periphylla are modified interradial tentacles, which function simultaneously as
acoustic and as optical organs of sense ; in some respects they appear allied more with the
sense clubs of Chart/odea, in other respects with those of Nausithoe. In our
species there are probably three small eyes furnished with pigment, lens and nerves

70 THE VOYAGE OF H.M.S. CHALLENGER,

above the auditory club ; of these the unpaired (lower) eye looks inwards, the two paired
(upper) eyes outwards.

The nervous system of PeriphyUa, like that of all the Peromedusse, is at present
unknown, and, unfortunately, in spite of repeated efforts, I was unable to make it out from
the single spirit-specimen examined. It requires fresh researches on living and specially-
prepared material. But considering the high stage of differentiation and perfection to
which the formation both of the muscular system and of the sense organs of our highly
developed Medusa has attained, we may assume that the nervous system is also fully
developed. This supposition is further justifiable from the fact that the closely-allied
Cubomedusse have a highly developed nervous system with centralised nerve ring, and that
the organs of sense show many analogies in the two orders. A nerve ring probably runs
in the coronal furrow as an important central organ, in immediate connection with the
four interradial sense clubs. A second nerve ring perhaps exists at the margin of the
coronal muscle, and possibly a third at the oral margin or the palatine ring. From the
large size of this Medusa, these important conditions might be explained by examination of
more perfectly preserved PeriphyUa treated with osmium and other such reagents.

The subumbral umbrella cavity (" antrum," PI. XIX. fig. 6 ; PL XX. fig. 8 ; PI.
XXI. figs. 12-19) in our PeriphyUa, as in all Peroniedusse, is divided into two distinct
sections, the distal simple coronal cavity and the proximal cpiadrilocular funnel cavity,
the palatine ring forming the boundary of the two. The distal (lower or oral) coronal
cavity of the umbrella (" antrum coronare," fig. 19, he) is simple, shaped on the whole like
a hemisphere or truncated cone, and enclosed round by the corona of the umbrella ; it is
7 cm. in height by 12 cm. in diameter, opens below to the outside by the opening of the
umbrella and contains the oesophagus in its centre. The upper boundary of the coronal
cavity against the funnel cavity is formed by the palatine ring (" annulus palatinus," wp).
I give this name to the important subumbral boundary ring between the cesophagus and the
coronal sinus, in whose plane the four perradial palatine nodes are inserted into the wall
of the sinus. Four wide horizontal openings, the funnel openings (" ostia infundibularia,"
fig. 18, if), leading from the coronal opening of the umbrella into the four interradial
funnel cavities (ii), are placed between the four palatine nodes.

The funnel cavities (" infundibula ") are conical ectodermal invaginations of the
subumbrella into the central stomach. They correspond completely to the "funnel
cavities " of the Lucernaridae, but are much more strongly developed, and play a much
more important part. Whilst in Pericolpa and Perrpalma they only reach the boundary
of the central stomach and basal stomach (as far as the pyloric ring), in Pericrypta and
l'< riphylla they completely hollow out the conical tseniola and also pass above into the
basal stomach as far as its conical point, where the csecal ends of their cones touch in the
subumbral centre of the umbrella cone. Each infundibulum represents a sub-regular cone
3 cm. high and 4 cm. in diameter at the base, and is divided by the horizontal boundary

REPOKT ON THE DEEP-SEA MEDUSAE. 71

line of the pyloric opening (gy, figs. 12, 13), into two distinct sections of equal height,
a lower distal half, and an upper proximal half. The lower or distal half of the funnel
(or the " central funnel," fig. 16, 17, ic) lies in the outer surface of the central stomach (gc) ;
its inner or axial wall is formed by the obelisk plates (yz) of the central stomach, its outer
or abaxial wall by the subumbral wall of the coronal sinus (cs). The upper or proximal
half of the funnel (or of the " basal funnel," fig. 14, ib) is encircled by the four niches of the
basal stomach (gn), and is only joined to the inner wall of the gelatinous umbrella at the
interradial line (fig. 14, ug). The four funnel cavities of the Peromedusas are homologous
with those of the Stauromedusaa and the Cubomedusas, and may also be comparable to the
subgenital cavities of the Discomedusas, the respiratory cavities of older authors on the
medusas. In fact, they may serve both as means of respiration and locomotion, as they
are emptied by each systole of the umbrella and filled with fresh water by each diastole ;
their wall is firm, but very thin.

The inner concave umbrella wall or subumbrella, shows a highly developed system
of strong swimming muscles, evolved from the more simple muscles, which I distinguished
in the Stauromedusee as the distal coronal muscle and the proximal bell muscle (comp. my
System der Medusas, 1879, pp. 366, 382, 399, 456, taf. xxi. xxx., &c). The coronal
muscle ("musculus coronaris," mc) is improved into a powerful broad band, and more
strongly developed than in all other Acraspedse. It consists of powerful leaves of the
circular muscle, whose thick supporting plate rises above the subumbral surface in the form
of 10 to 12 strong circular folds ; the height of these circular folds (?nc2) decreases from
above to below, their breadth increases (PI. XIX. fig. 6 ; PI. XXI. fig. 8 ; PL XXII. fig. 22,mc).
The upper or proximal margin of the coronal muscle (figs. 8, 22, mc-i) forms a simple cir-
cular line, corresponding with the distal margin of the large circular sinus. The lower or
distal margin of the coronal muscle (figs. 8, 22, mc4) forms sixteen triangular, subradially
projecting points which run as far as the middle of the marginal lobes. The whole
coronal muscle is therefore divided by sixteen selvages into sixteen quadrangular area?,
the coronal plates ("tabulae coronares"). The four interradial ("ocular") are some-
what narrower than the remaining twelve (" tentacular "). They are divided by the lobe
clasps (" loboporpss," M), the longitudinal fused selvages, which divides each marginal
lobe into two halves (comp. pp. 66, 67) ; they serve at the same time as firm cartilage-
bke selvages for the insertion of the circular muscular fibres. Each quadrangular coronal
plate corresponds, therefore, to the adjacent halves of two marginal lobes, and connects
them most closely. Four of the muscular areas correspond at the same time to the four
interradial sense clubs, whilst the twelve others correspond to the tentacles. The two
lateral margins of each muscular area (formed by the lobe clasps) are straight, and con-
verge upwards ; the lower or distal margin is the larger, and sinuated concavely ; the
upper or proximal margin is convex, and touches alternately the basis of the deltoid
muscles and the distal end of the genitalia.

72 THE VOYAGE OF H.M.S. CHALLENGER.

Whilst the large coronal muscle with its circular fibres contracts the distal part of
the suburnbrella, the muscle of the swimming bell (" musculus codonoides ") with its
longitudinal fibres answers for the proximal half of the suburnbrella. The most import-
ant longitudinal muscles of this system are the eight strong deltoid muscles ("musculi
deltoides," md ; PI. XIX. fig. 6 ; PL XX. fig. 8). They are very powerful, ecmflaterally
triangular, and touch the proximal margin of the coronal muscle with their broad bases,
whilst their truncated point is directed upwards and their longitudinal fibres consequently
diverge centripetally. The four weaker perradial deltoid muscles (md') are inserted by
their truncated point at the distal end of the gastral openings, in the subumbral wall of
the four cartilaginous palatine nodes (gh). The four stronger interradial deltoid muscles
{md"), on the other hand, are longer, and inserted further up on the subumbral wall of the
four septal nodes (hi) in the middle of the length of each pair of genitalia between the two
halves. Between these, the deltoid muscle also forms, above the septal node, a thin band-
shaped prolongation, which runs centripetally as far as the pylorus ("musculus intergenitalis,"
fig. 8, ms). Besides these, a stronger longitudinal muscle, which I wdl call "musculus
congenitalis" (fig. 8, mn), runs into the suburnbrella on the two lateral margins of each of
the four gastral openings, between them and the limiting genital bands. It springs with
a broader base from the inverted lateral margin of the perradial deltoid muscle, runs,
gradually becoming narrower, up above between the gastral ostium and the limbs of the
genitalia, and is inserted above with its narrow end in the pyloric ring (fig. 8, gy).
Finally, a narrower and very much weaker longitudinal muscle, which may be termed
" musculus axogenitalis," runs in the middle of the eight genitalia, and, in fact, on the
midrib between the two limbs of each genitalium (fig. 38, mx). On the whole, therefore, the
system of the muscle of the swimming bell (" musculus codonoides ") is divided into four
stronger interradial and four weaker perradial area? ; the four interradial deltoid muscles,
the four intergenital muscles and the eight axogenital muscles belong to the former ; the
four perradial deltoid muscles, and the eight congenital muscles to the latter.

Although the circular system of the distal coronal muscle, and the longitudinal system
of the proximal swimming bell muscle, form by far the most important part of the
subumbral muscidar system, it is represented by weaker muscles in other parts of the
suburnbrella. The circular fibres already mentioned, in the wall of the basal funnel
cavities, belong to it on the one hand, and the longitudinal fibres on the concave axial side
of the marginal lobes, which we shall briefly term " lobe muscles " (" musculi lobares," mh),
on the other. Each of the sixteen marginal lobes has in its thin subumbral wall a pair of
such longitudinal muscular bands, which run to both sides of the medial (subraclial) lobe
clasps (Ik), and clearly correspond to the well-known stronger lobe muscles of the Disco-
medusse.

The gastro-vascular system of Periphylla mirabilis (PL XX. figs. 8-11; PL XXI.
figs. 12-20 ; PL XXII. fig. 22 ; PL XXIII. figs. 29-31) is distinguished from that of all

REPORT ON THE DEEP-SEA MEDUSAE. 73

Peromedusse by many very peculiar and complicated arrangements, which can only be
completely understood after long and minute study. The only gastrovascular system
among the forms of the Medusae systems hitherto known, which offers more detailed
points of comparison, is that of the Lucernaridse, and in PervphyUa, as in the Lucernaridse,
it is derived from that of Tesseridse (comp. the general anatomical representation of the
Stauromedusse in my System, 1879, pp. 363-395, taf. xxi., xxii.). But whilst the
essential condition of the central gastrovascular system in those Stauromedusae
resembles that of the Perornedusse, in the Peromedusae it differs in detad in very import-
ant and peculiar complications, and especially in the formation of the peripheric part.
Generally speaking, we can distinguish in all Peromedusas two principal sections of the
gastrovascular system, the central principal intestine (" gaster principalis ") and the
peripheric coronal intestine (" gaster coronaris ") ; they are only connected by four
narrow, cleft-shaped, perradial gastral openings, and are otherwise completely separate.

The central principal intestine (" gaster principalis ") occupies the entire length of the
axial space of the body and extends from the umbrella cone to the oral margin. It is
divided by two circular constrictions or horizontal strictures into three principal sections
— basal stomach (gb), central stomach (gc), and buccal stomach (ga). The upper or aboral
circular stricture between the basal and the central stomachs, I term the pyloric opening
or pylorus (" porta pylorica," gy) ; the lower or oral circular stricture, between the
central and the buccal stomachs, the palatine opening or palatum (" porta palatina," gj)).
In order to gain a general correct idea of the complicated conditions of form of these three
sections of the principal intestine, it is convenient to refer them to simple mathematical
figures ; the basal stomach is a cone (or more properly a epiadrate pyramid), the central
stomach an obelisk (or a truncated, regular quadrangular pyramid), the buccal stomach
a cube (or a four-sided prism) . The central stomach communicates with the peripheric
circular sinus of the coronal intestine by four lanceolate perradial openings (" ostia
gastralia ").

The buccal stomach (eventually also termed proboscis or oesophagus, "gaster
buccalis, tubus oralis, proboscis " PL XIX. fig. 6 in the centre ; PI. XX. fig. 1 1 ;
PI. XXI. fig. 19) forms the lowest oral third of the axial principal intestine, is dice-shaped
on the whole, and hangs perfectly freely in the centre of the coronal hollow of the umbrella,
as it is only fastened above to the suburnbrella by the four perradial palatine nodes (cfk).
The length of the edge of the cube amounts to nearly 5 cm.; Plate XX. shows it apart
in its natural size, fig. 9 the interradial external view, fig. 10 the perradial external view,
and fig. 11 the perradial transverse section. The lower (oral) wall of the cube occupies
the quadrate oral opening (oa), and the upper (aboral) wall the palatine opening (gp) ;
the four perradial vertical lateral surfaces of the cube are formed by the buccal pouches
(bb), the four interradial edges between them by the buccal columns (ac).

The oral opening (" osculum," cm), is shown from below in the middle of fig. 6 of

(ZOOL. CHALL. EXP. — PART XII. — 1881.) M 10

74 THE VOYAGE OF H.M.S. CHALLENGER.

Plate XIX. ; it forms a quadrangle with rounded angles. The delicate, thin membranous
margin of the mouth (oral margin) is contracted inside like a narrow velum, and appears
swollen and thickened at the four interradial angles by the crescentic oral ends of the
buccal columns which are concave inside. Each of the latter bears two thin oral filaments
two cm. long at the end of the horns of the crescent (" barbulse, filamenta oralia," figs. 9-1 1 ,
«/). These may probably be considered as the last oral branches of the limbs of the
tasniola (see below). They are thickened conically at the base, and run out to the point
in a very thin long filament (or in a pencil-shaped bunch of filaments) ; they are amply
furnished with large bean-shaped nematocysts, whose urticating threads are twisted
spirally and armed with bristles.

The oral cavity (" cavitas buccalis ") is divided by the four interradial buccal columns
into four perradial peripheric buccal pouches (bb), which only communicate with the cen-
tral cavity of the mouth (ax) by four narrow oesophageal clefts. The four oral columns
("columnse buccales," ac, figs. 9-11, 19) are nearly rectangular ridges or plates, 5 cm.
hio-h and 2-3 cm. broad, projecting inwards in the interradial meridian plane into the oral
cavity. They are supported by a visible layer of gelatinous substance, several millimetres
thick, which is thickest at the two lateral margins and in the middle of each plate, so that
each plate is also traversed on its gastral surface by a pair of shallow, parallel, longitudinal
grooves (transverse section, fig. 19, ac). The lateral parts of the buccal columns project
like wings on the two sides of the groove (adradial oral wings, "ake buccales," ad).
In this respect they resemble the taeniola of the Scyphostoma, and, in fact, I consider them
homologous with the peristome part of the latter. In Periphylla mirabilis, moreover,
they are much less strongly developed than in the following species : — Perip>hylla regina
(PL XXIV. fig. 3) and Periphylla hyacinthina (System, 1879, taf. xxiv. fig. 14). The
four perradial egg-shaped buccal pouches ("bursas buccales," bb, figs. 9-11, 19) project arch-
ing out externally between the buccal columns. The central spaces only of each buccal
pouch opens freely into the oral cavity, then- peripheric spaces have dilatations or horns
which are covered for the most part by projections of the enclosing wall. Each buccal
pouch is therefore divided by the projecting wings of the buccal columns into the open
central space, and the lateral horns or wing pouches (" ventriculi laterales, bursas alares," bd)
covered by the buccal columns. Each side pouch passes above into a larger and deeper
aboral corner horn, below into a smaller and shallower oral corner horn ; the former ends
caseally in the upper thickened end of the wing of the buccal column (fig. 11, ad). The
corner horns are not so depressed in this species as in the following one. The four perradial
buccal clefts ("fissuree buccales," ae), by which the four buccal pouches communicate with
the central space of the oral cavity, are narrowed in the middle. The buccal pouches are
inflated ovally out from them (figs. 9, 10, bb). The perradial wall of the buccal pouches
is very much thinned, and is traversed by parallel longitudinal streaks, which are
divided by fine transverse streaks into darker cubes (oral glands, fig. 10, ag). This wall

REPORT ON THE DEEP-SEA MEDUSAE. 75

becomes considerably thickened above, and then passes directly into the four perradial
palatine nodes (gk), by which the buccal stomach is fixed to the subumbrella.

The palate or palatine opening ("palatum, porta palatum," #p ; PI. XX. fig. 11 ; PI. XXI.
figs. 12, 13, 18, gp) forms the important opening for communication between the buccal
stomach (ga) and central stomach (gc) ; it can probably be completely closed by muscular
contraction in the living Periphylla. Properly speaking, it consists of the wide central
palatine opening and the four perradial palatine grooves surrounding it. The central
palatine opening (" porta palatina", gp) is quadrate, its interradial lateral margins are
formed by the upper, swollen and thickened, aboral margins of the buccal columns (ac),
which here pass immediately into the lower delicate oral margins of the thin obelisk
plates (gz). Its perradial corners, on the other hand, communicate by a narrow cleft
(perhaps capable of closing) with the four palatine grooves which represent in some
measure four secondary openings of the central principal opening (figs. 8, 11, 18, ga).
These four perradial palatine grooves (" sulci palatini," gs) lead from the oral cavity
immediately towards the outside into the coronal sinus, and form, at the same time the
dilated distal ends of the cleft-shaped gastral openings (go). They are imbedded in the
cartilaginous mass of the four palatine nodes (" nodi palatini," gh). It appears that the
palatine grooves remain open even when the principal opening is completely closed,
and then by contact of the two lips of their fissure can be transformed into short closed
canals (of about 3 mm. in diameter).

The central stomach (" gaster centralis, obelisk stomach," gc ; PI. XX. fig. 8 ; PI.
XXI. figs. 11-18), the middle of the three divisions of the axial principal intestine, is
somewhat smaller than the buccal stomach, and has, on the whole, the form of an obelisk
or a truncated regularly quadrilateral pyramid (figs. 12, 13, gc). We can distinguish
geometrically two bases and four lateral surfaces. The lower (oral) base is formed by the
palatine opening described above (" palatum," gp), by which the central stomach opens
into the buccal stomach. The upper (aboral) basis, on the other hand, occupies the
quadratic pyloric opening (" pylorus" gy), by which the central stomach communicates
with the basal stomach. The four interradial lateral surfaces of the obelisk-shaped central
stomach form four trapezoid, or almost rectangular thin lamellae, which on account of their
special importance I have termed (once for all, to prevent confusion) the four obelisk plates
of the central stomach (" tabulae obelisci," gz). The thin wall of these quadrangular plates,
which are placed more or less vertically, belongs properly to the subumbrella, and is
formed by a delicate but firm gelatinous plate or supporting lamella, whose inner or
axial surface is covered by gastral endoderm and its outer or abaxial surface by the
subumbral ectoderm of the funnel cavities, and a thin layer of muscle belonging to it.
The upper or aboral margin of each obelisk plate is formed by a quadrant of the pyloric
stricture (gy), the lower or aboral margin by a quadrant of the palatine stricture (gp),
whilst the two lateral or longitudinal margins are beset with a row of gastral filaments

76 THE VOYAGE OF H.M.S. CHALLENGER.

(fe), and form the lateral boundary margin of a gastral opening (go). The four gastral
openings or perradial clefts of the central stomach (" ostia gastralia," go), are four wide, long
cleft openings, by which the central stomach communicates in its whole length with the
coronal sinus, and whose middle line corresponds to the four perradial borders of the obelisk
or, what is the same thing, to the ideal boundary line, at which each of the two obelisk
plates touch. The gastral openings have a narrow lanceolate shape, and are broadest in the
middle (6-8 mm.) and 36-40 mm. long (figs. 8, go; 12, go). The upper or aboral pointed
end of each gastral opening touches the perradial point of the pyloric stricture (gy) ; on
the other hand, the lower or oral end touches the perradial point of the palatine opening
(gp), and is intersected here in the form of a peculiar groove, the palatine groove already
described, which is embedded in the firm palatine nodes. The border of gastral filaments
(fg) ends somewhat above the palatine groove.

The pylorus, or pyloric opening, is the name which I have given to the quadrate
opening, by which the central stomach communicates with the basal stomach (" pylorus,
porta pylorica," gy ; figs. 8, 12, 13 ; gy, fig. 15). The four perradial angles of this quad-
rate form the aboral end of the four gastral ostia (go). The four interradial lateral fines
of the quadrate, 3 cm. in length, form the upper (aboral) boundary lines of the obelisk
plates, in which they touch the axial walls of the basal funnel cavities. As each of the
two adjacent basal funnels touch each other by their lower ends at the pyloric opening, two
gastral filaments (fig. 15, b) are placed in each corner of the pyloric quadrate. It is
only in the four pyloric corners (fig. 15, gy) that the wall of the pyloric opening touches
the gelatinous wall of the umbrella (ug), from which it is otherwise completely separated
by the four interradial subumbral funnel cavities (ii).

The basal stomach (" gaster basalis," peduncle stomach, gb ; PL XX. fig. 8 ; PL XXI.
fig. 14) forms the upper or aboral third of the axial principal intestine, and has the
form of a regular hollow cone, whose base is the pylorus, and whose point is the cone of
the umbrella. As, however, it encloses the four interradial conical funnel cavities, it really
has the geometrical fundamental form of a quadrilateral regular pyramid. This is 4 cm.
high, whilst the length of its edges amounts to 5 cm. and the lateral length of its quad-
rate base to 3 cm. The point of the pyramid is prolonged into a narrow caecal tube,
which traverses the aboral cone of the gelatinous umbrella, and whose point nearly
touches the external surface of the latter. This canal of the umbrella peduncle- (fig. 8, cb)
is here closed caacally, and does not open by an aj>erture into the upper surface, as
appears at first sight. The cavity of the narrow spindle-shaped peduncle canal is fined with
dark-brown pigment, and therefore stands out conspicuously in the clear gelatinous mass
(fig. 1, in the point above). As the four interradial conical subumbral funnel cavities (ib)
already described traverse the whole length of the basal stomach and meet above in its
point, the periphery of its conical hollow space is divided into four perradial grooves, the
basal pouches, or niches of the basal stomach (" bursas basales," gn). They are

REPORT ON THE DEEP-SEA MEDUSJE. 77

broadest in the middle, lanceolate above and below (figs. 8, 12, gn). They communicate
with the simple central space of the basal stomach by four narrower longitudinal clefts,
and correspond to the four peduncle chambers of Lucernaria (Pis. XVI., XV 1 1., gn).
The transverse section is therefore the same in both cases, and shows the form of a
Maltese cross ; but with this difference, that in Lucernaria (PI. XVII. fig. 13) as in Peri-
colpa and Peripalma the four interradial tseniola are solid selvages, whilst in PeriphgUa
(PI. XXI. fig. 14) as in Periphema and Pericrgpta they are hollow cones. Two diverging
phacelli (or longitudinal rows of gastral filaments) beset the entire length of the hollow
cone, and diverge from each other below at the pylorus, in such a way that the two
phacelli of each two adjacent cones which are turned to each other meet in the four
perradial angles of the pylorus. From thence they diverge further upon the margin of
the gastral openings.

The tseniola ("tasniola gastralia," gastral longitudinal selvages, ft). The axial princijml
intestine of PeriphgUa, whose three divisions have been already described, has apparently
an extremely complicated character, which separates it in a striking manner from
other Medusas. A clear, simple explanation of this may, however, be gained by com-
paring this axial intestine with the more simple principal intestine of the Lucerna-
rida3 and Tesseridse. If we abstract the secondary differentiations, and only bring
forward the primary principal conditions, we are able to refer all these formations
to the simple, common ancestral form, to the primary intestine of the scyphopolyps,
Scgphostoma (comp. my System der Medusen, pp. 364, 3G7, 384, 403, &c). The four
endodermal interradial tseniola are already developed from this primary intestine, divide
the periphery into four perradial niches or pouches, and traverse the whole length of the
gastral wall, from the aboral peduncle base to the oral margin. These then charac-
terise pre-eminently the section of the Acraspedas, and develop the peculiar typical
" gastral filaments." In their common parent form, Tessera as in Scgphostoma, we can
distinguish two sections in each tseniolum, the umbral at the umbrella wall and the sub-
unibral at the peristome wall ; the two touch at the umbrella margin. From beginning to
end, from the aboral central point to the oral margin, the interradial tgeniola and their
products show a steady tendency to centripetal growth, whilst on the contrary the
perradial pouches between them show the same tendency to centripetal growth. In our
PeriphgUa (1) the four funnels of the basal stomach and their rows of filaments, (2) the
obehsk plates of the central stomach with rows of filaments, (3) the buccal columns of the
buccal stomach with their wings and oral filaments belong to the centripetal system of
the four interradial taeniola. On the other hand, (1) the four niches of the basal stomach,
(2) the gastral openings of the central stomach leading into the peripheric coronal
intestine, (3) the buccal pouches and wing pouches of the buccal stomach belong to the
centrifugal system of the four perradial pouches. The correctness of this view is proved
directly by the distribution of the eight phacelli or rows of filaments, of which each two

78 THE VOYAGE OF H.M.S. CHALLENGER.

come on one taeniolum, and are placed in some measure on two diverging limbs of the
tseniola.

The phacelli or the longitudinal rows of the gastral filaments (PI. XX. fig. 8 ; PI.
XXI. figs. 14-18 ; PI. XXII. figs. 23-28) are extremely powerfully developed both in
PeriplujUa mirdbilis and in the following Periphema regina (PI. XXIV. fig. 1). The
number of filaments amounts to several thousand, and their length to 30 or even 40 mm.
They are apparently distributed over the whole extent of the basal and central stomach
that they form eight continuous longitudinal rows or " phacelli," which run divergingly
from the conical basal stomach. Closer inspection, however, shows that the two phacelli
of each pair originate as diverging limbs, from a simple interradial phacellus deep in the
bottom of the basal stomach. They there form a simple row of short filaments, which
stand freely on the interradial taeniola and project into the basal gastral cavity. This
simple phacellus soon divides into two limbs, which diverge only slightly at first but
more strongly afterwards. At the pylorus they diverge so strongly that they touch the
meeting limbs of the adjacent tamiola in the four perradial angles of the pylorus. They
then run along the margin of the gastral openings (go) nearly to the upper margin of the
palatine groove. Each perradial gastral opening is bordered on both sides of the
margin by a row of long gastral filaments, which project freely into the central gastral
cavity. These filaments are generally 1-2 cm. long. ; many of them, however, 3-4 cm.
long ; their breadth varies between £ and 1 mm., but often amounts to 1^-2 mm.
They are sometimes more cylindrical in shape, sometimes flattened like a ribbou, often
tongue-shaped at the end (PI. XXII. fig. 23 ; transverse section figs. 24, 25). The struc-
ture of these gastral filaments is the same as usual (fig. 26). A gelatinous supporting
plate (z), enclosing scattered cells, is covered with an endodermal epithelium, which con-
tains three kinds of cells, (1) narrow, high, cylindrical flagellate cells (fe) ; (2) flask-
shaped glandular cells with turbid contents, consisting partly of finely granular protoplasm,
partly of large, strongly-refractive corpuscles (fd) ; (3) thread-shaped epithelial muscular
cells containing nuclei (fm). These endodermal muscular cells, hitherto sought for in
vain, exist, I believe, isolated here in the large contractile and very mobile gastral fila-
ments (fig. 28).

The peripheric coronal intestine (" gaster coronaris ") includes the entire peripheric
part of the gastrovascular system (as opposed to the axial principal intestine) and
occupies the whole subumbrella from the pylorus to the umbrella margin. It is divided
into two principal sections, which are separated by the upper or proximal margin of the
coronal muscle. The upper or proximal section itself fills the large coronal sinus, whilst
the lower or distal section forms the peripheric corona of pouches. This consists of sixteen
quadrangular coronal pouches, which correspond to the coronal plates of the coronal
muscle. Three pouches, two lateral lobe pouches, and a middle pouch passing into a
tentacle or a sense club, run out from the distal margin of each coronal pouch. The

REPORT ON THE DEEP-SEA MEDUSAE. 79

peripheric coronal intestine is only connected openly with the axial principal intestine
at four points, viz., at the four perradial gastral openings (go).

The remarkable, enormously large circular sinus, or coronal sinus (" sinus coronaris,"
or " canalis coronaris,") cs ; PI. XX. fig. 8, cs ; PI. XXI. figs. 12-18, cs) is that part
of the gastro vascular system which specially distinguishes the Peroniedusee from all
other Medusae, and which does not recur in the same form and to the same extent in
any other group of Medusas. It forms a colossal annular pouch which is placed more or
less subvertically, and in Pervphylla mirabilis is 30 mm. high, so that it fills the half
of the whole height of the umbrella. In the circular sinus we distinguish an upper or
proximal and a lower or distal margin, and an inner or axial and an outer or abaxial
wall. The upper margin and the outer wall of the circular sinus are without any
opening ; on the other hand, it communicates at the lower margin by sixteen horizontal
clefts, with the sixteen coronal pouches, and at the inner wall by the four vertical gastral
openings with the central stomach. The lower or distal margin (fig. 15, cs) coincides
with the proximal margin of the coronal muscle (mc), and has, consequently, sixteen sub-
radial projecting corners (PI. XIX. fig. 6). The aboral or proximal margin (fig. 15, cs),
is a simple circular ring of the subumbrella, which coincides with the plane of the
pylorus ; the hollow space of the coronal sinus is here completely closed, and does not
communicate with the surrounding circular pylorus ; the proximal margin of the circular
sinus (fig. 15, cs) is completely separated from the pyloric opening (gy) by the four
interradial funnel cavities (ii) (which touch here), and only touches the pyloric opening
externally in the four perradial pyloric corners (gyl). The external, umbral, or abaxial
wall is formed by the smooth concave inner surface of the gelatinous umbrella, and is
covered by dark pigment, which is deposited in the form of black-brown balls in the
endoderm cells of the umbral wall. The inner, subumbral or axial wall of the circular
sinus is formed by the subumbrella, which here assumes very compbcated conditions
in consequence of the strong development of the four funnel cavities. Considered more
closely, this axial wall is divided into an upper broader, and a lower narrower section,
which is bounded by the subumbral palatine ring or the insertion of the four perradial
palatine nodes (gh) at the subumbrella. The upper or proximal section of the axial
wall, above the palatine nodes, is pierced in its entire length by the four perradial gastral
openings, these important clefts already described, by which the hollow space of the
central stomach opens into the coronal sinus (go). The lower or distal section of the
axial wall, on the contrary, is perfectly simple, without openings ; the deltoid muscles
(md) lie on its subumbral surface. Moreover, the four pair of genitalia (fig. 20, sm) lie in
the four quadrants of the axial wall of the coronal sinus, which are separated by the four
gastral ostia, The colossal hollow space of the coronal sinus forms a powerful closed
pouch; apart from the four septal nodes which I shall next describe, it is perfectly simple,
and, moreover, so wide that I could easily introduce three of my fingers, and so realise

80 THE VOYAGE OF H.M.S. CHALLENGER.

its full extent. The powerful volume of its contents is probably subject to important
modifications according to the state of contraction of its muscular and very extensible
subumbral wall.

The inner or axial wall of the coronal sinus is firmly connected with its external or
abaxial wall at four interradial points. These points lie exactly in the radii of the four sense
clubs, in the middle third of the height of the sinus, a little above the middle (figs. 8, 12,
13, hi). They are the important septal nodes or fused nodes (" nodi septales, nodi
cathammales," hi; PL XX. fig. 8, hi; PI. XXI. fig. 17, hi; PI. XXIII. fig. 33;
PL XXV. fig. 8) ; they correspond to the interradial septa of the Lucernaridse and
Cuboruedusaa, which divide the length of their radial pouches. Each of the septal nodes
forms a gelatinous cube, tiny but firm as cartilage, of from 2-3 mm. in diameter. Under
strong magnifying power, the transverse section (PL XXV. fig. 8) shows that the
umbral gelatinous substance of the thick abaxial wall (wg) is firmly fused here with
the gelatinous supporting plate of the subumbral axial wall (ziv). Between the two
walls, however, in the tangential median plane of each node, there is a double layer of
endoderm cells, the important cathammal plate (" vascular plate or endodermal lamella,"
dk). The higher cylindrical cells of the axial cell layer (dw.2) form the direct continuation
of the subumbral endodermal epithelium of the inner sinus wall (dw) ; in the same way the
flatter cells of the abaxial cell layer (du2) pass immediately into the umbral endodermal
epithelium of the external sinus wall (du), and contain the black-brown round pigment
granules by which the wall is characterised. We cannot, therefore, doubt that the nodes
are really formed by fusion of the two vascular walls. The gelatinous substance on
both sides of the double gastral lamella was so much hardened that it notched the knife
in making sections. Under stronger magnifying power (PL XXV. fig. 8) it showed
here the same striking induration and histological modification of the gelatinous tissue,
already described in the periphery of the lobe clasps, the transition into hard fibrous
cartilage (comp. above p. 67, and PL XXV. fig. 10).

If we suppose that the four septal nodes, prolonged centripetally to the pylorus and
developed into four selvages, connect the axial and abaxial wall of the coronal sinus in
its entire proximal half, the sinus would thereby be divided into four wide perradial
pouches, corresponding to those of the Lucernaridse and Charybdeidae ; and as in these the
four radial pouches here communicate below by a circular canal, viz., by the simple distal
half of the circular sinus, below the septal nodes. We may therefore say the ideal
horizontal circular line, in which the four interradial sejotal nodes lie, form the boundary
line between an upper and a lower coronal sinus, both of which are in open communica-
tion between the four nodes. The upper or proximal coronal sinus (cs4) corresponds to four
wide perradial pouches, whose septa are reduced to nodes, and which communicate by
the gastral openings with the central stomach. The lower or distal coronal sinus (cs.2)
corresponds to a very much widened coronal canal which connects the four radial pouches

REPORT ON THE DEEP-SEA MEDUSAE. 81

at the umbrella margin (as in the Lucemaridse and Chary bdeidas). The correctness of this
morphological view is also justified by comparison with the important common parent
group of the Tesseridse (System, p. 369, taf. xxi.).

The large coronal sinus is divided by the upper or proximal margin of the coronal
muscle from the marginal pouch corona, which forms the principal section of the
peripheric coronal intestine ("corona bursarum," PL XXI. figs. 12, 13, 19, 20 ; PL XXII.
fig. 22 ; PL XXIII. figs. 29-32 ; PL XXIV. fig. 1). This corona is composed of the sixteen
coronal pouches and the canals which run from them into the four sense clubs, the twelve
tentacles, and the sixteen marginal lobes. The sixteen coronal pouches ("bursas coronares,"
be) into which the lower or distal margin of the circular sinus opens (at the proximal
margin of the large coronal muscle), correspond in number, shape, and size to the sixteen
coronal plates or the separate trapezoid muscular areas of the large coronal muscle (me).
They are shallow quadrangular pouches, whose inner or axial wall is formed by the
folded muscular area itself, its outer or abaxial wall by the smooth internal surface of the
gelatinous umbrella on whose external surface there is a pedalium corresponding to each
coronal pouch. The upper or proximal margin is formed by the horizontal narrow
cleft, by which it communicates with the coronal sinus ; it corresponds to the subumbral
boundary line between the coronal muscle and deltoid muscle. The two lateral (or
radial) margins are formed by the lobe clasps (M), by which each coronal pouch is divided
all its length from the two neighbouring pouches. As each lobe clasp cuts a marginal
lobe all its length into two halves, each coronal pouch belongs to the adjacent halves of
two lobes and sends out an evagination, the lobe pouch (" bursa lobaris," or lobe canal,
" canalis lobaris," fig. 22, bl ; fig. 29, bl) into each of these halves. As the lobe clasp
(M) only halves the upper or proximal part of the lobe and leaves the lower or distal
part free, both pouches of each lobe are in open communication below the clasp. They
consequently form a horseshoe-shaped canal, whose two parallel limbs are directed
centripetally and only separated by the septum of the lobe clasp (horseshoe canal, "bursse
hipposideri," fig. 22, bw ; fig. 29, bu). its proximal openings are in two adjacent
coronal pouches. If we fill one of the two lobe pouches of a coronal pouch with air, the
air passes through the U-shaped canal into the adjacent coronal pouch (fig. 22, bu). In
this way there actually arises in all Peromedusae a connective circular canal at the
umbrella margin, which in some measure resembles the festoon canal of the Narco-
medusse, runs along the margin of all the lobes, and puts all the coronal pouches into
peripheric communication. In the Pericolpidse, this wide festoon canal or marginal
canal (" canalis marginalia," cm) is composed of eight coronal pouches and sixteen lobe
pouches, whilst in the Periphyllidse it is composed of sixteen coronal pouches and thirty-
two lobe pouches (comp. my System cler Medusen, taf. xxiii. xxiv.).

As the four interradial areas of the coronal muscle corresponding to the sense clubs are
considerably narrower than the twelve remaining areas corresponding to the tentacles, the

(ZOOL. CHALL. EXP. — PART XII. — 1881.) M 11

82 THE VOYAGE OF H.M.S. CHALLENGER.

same of course holds good of the coronal pouches whose subumbral wall forms the folded
muscular arese. In the middle of the lower or distal margin of each coronal pouch,
just where its two lobe pouches opens into it, a canal also runs out from it between the
two lobe pouches which leads into the tentacle inserted in the coronal pouch. The four
interradial sense canals ("bursse sensillares," bo) which provide for the four sense clubs, are
short and simple, and swell into a spheroidal vesicle (" ampulla rhopalaris," oa ; PI. XIX.
figs. 2-3 ; PI. XXII. fig. 22 ; PL XXIII. figs. 31, 32, oa) at the basis of each sense club (on
the axial side). The formation of the twelve tentacle canals (of which four are perradial
and eight adradial) is more complicated. At the tentacle basis, below the two tentacle
roots, these canals can be closed by the peculiar double valvular vent-hole already de-
scribed (comp. p. 68, and PI. XXII. fig. 22, yk' , ex).

These complicated anatomical conditions of the peripheric pouch corona are more
difficult to understand, inasmuch as each of the twelve tentacular coronal pouches (but not
the four ocular coronal pouches) are divided into two pouches by an imperfect tangential
septum (PL XXII. fig. 22 ; PL XXIII. fig. 29). These two pouches, the inner or axial
velar pouch (be), and the outer or abaxial avelar pouch (be'"), communicate by a lon-
gitudinal cleft in the middle of the septum which divides them ("fissura septalis," be'").
This peculiar complication arises from each tentacle sending out above at its insertion
(between two marginal lobes) two diverging centripetal muscles, the root muscles of the
tentacles (mk) already described. These invaginate the lower or distal margin of the
coronal pouch in such a way that each tentacle root is surrounded by a conical
ectodermal hollow space, the funnel cavity of the tentacle root (it). The cascal end of
this funnel cavity extends to the upper or proximal margin of the coronal muscle where
the point of the tentacle root is inserted. The septal fissure, by which the axial velar
pouch communicates with the abaxial avelar pouch, remains between the two bifurcate
diverging tentacle roots (mk). The "septum velare" (wm), which itself is hollow and
separates the two pouches, has consequently a very complicated formation. It is formed
by two parallel lamellae of the velar fold, which only pass into one another above at the
proximal margin of the coronal pouches and at the two margins of the septal fissure.
The space between the two lamellae, the funnel cavity of the coronal pouch (" infundibulum
coronare," ic) is lined by the ectoderm of the subumbrella, and divided into a distal
simple " funnel cavity of the tentacle base," and two diverging cEecal horns running
proximally from it, the two "funnel cavities of the tentacle roots" (it). The muscular
wall of the delicate membranes which separate these cavities forms part of the iu-
vaginated coronal muscle, and is laid in delicate folds, as is best seen from the figure of
the partially-opened coronal muscle in Plate XXIV. fig. 1 .

Genitalia ("sexualia," s ; PL XIX. fig. 6 ; PL XX. fig. 8 ; PL XXI. figs. 17 18 ;
PL XXII. figs. 38-40). The single specimen examined of Perijihylla mirabilis
was a mature male, whose testes had shed most of the spermatozoa. The testes

REPORT ON THE DEEP-SEA MEDUSAE. 83

("spermaria," sm) form eight horseshoe-shaped or U-shaped glands, which lie adradially in
the sulmmbral wall of the coronal sinus. The convex arch of each horseshoe is turned
distally, and nearly touches the proximal margin of the coronal muscle (mc), whilst the two
parallel limbs of the horseshoe are directed proximally (or centripetally), and their points
nearly touch the pyloric stricture (gy). The eight U-shaped reproductive glands therefore
extend in an adradial direction through nearly the entire height of the coronal sinus (os),
and only leave a small part of the upper (proximal) margin and the lower (distal) margin
of its subumbral wall free. The eight genitalia are connected in pairs, in such a way that
a pair lies between each two perradial gastral openings {(jo). The two genitalia of each
pah- are divided by the intergenital longitudinal muscle (ms) in the upper (proximal)
half, by the septal nodes (hi) in the middle, and by the interradial deltoid muscle (md) in
the lower (distal) half. The four pairs, on the other hand, arc divided by the four gastral
openings (go) in the upper half, and by the four perradial deltoid muscles (md) in the
lower half. The lower half only lies freely in the coronal cavity of the umbrella (he),
whilst the upper half is concealed deep in the funnel cavity (ie). The four limbs of each
pair run nearly parallel in the upper half (in the funnel cavity), whilst the two
genitalia of each pair diverge distally (in the coronal cavity of the umbrella), as they
diverge parallel to the two limbs of the interradial deltoid muscle. The two limbs of
each genitalium converge, however, at their distal end, and are connected there by the
cords of the horseshoe, whose convex outer margin nearly touches the inner margin of
the coronal muscle (comp. PL XX. fig. 8, and PL XXV. fig. 1). The two limbs of the
horseshoe-shaped genital band are broadest below (1 cm.), and gradually narrow as they
run up (0"5 cm.) towards the proximal points of the two limbs (PL XXIII. fig. 38).
The limbs are of ecpaal length in Periphytta mirabilis, whilst in Periphylla hyacinthina
the lateral limb (next the gastral opening) is much shorter than the medial limb
(next the septal nodes). (Comp. my System, PL XXII. figs. 13, 16.)

Structure of the spermaria (PL XXIII. figs. 38-48). Each of the eight horseshoe-shaped
genitalia shows a projecting, nearly adradial, selvage, the genital rib ("costa genitalis,
sterigma," st) between its two limbs. It consists of a firm selvage-like thickening of
the fulcra! plate of the subumbrella and of the weak axogenital longitudinal muscle
(mx) lying on it. Gelatinous transverse selvages run, as in a pinnated leaf, from both
sides of this midrib, and serve to support the separate pouch-shaped transverse folds of
the genitalium (fig. 38). More minute examination shows that the testis represents a
broad U-shaped arched band having many transverse folds along its whole length. The
convex lateral margin of the U-shaped band is fastened to the subumbral wall of the
ring sinus, whilst the concave, medial margin projects slightly into the canal space of
the ring sinus. The separate pouch-shaped, arched-out, transverse folds of the genital band,
amounting to about 50 to 60 in each genitalium, are of a narrow, oval or conical shape,
and bear the same relation to the midrib that the pinnae do in a pinnate leaf. Numerous

84 THE VOYAGE OF H.M.S. CHALLENGER.

secondary folds lie in each transverse fold. The interspaces between these folds perhaps
develop into special genital sinuses with excretory passages opening into the pouch
space and from there into the ring sinus, which probably comport themselves as in the
complicated forms of the Lucernaridaa. The complicated structure of this many-folded
genital band, is, however, as in the Lucernaridae, very difficult to make out. In the
single specimen of Periphylla mirabilis before me, the ripe pouches of the testes were
already flattened for the most part, and the spermatozoa emptied into the coronal sinus.
The small follicles of the testes which, closely placed together, compose the folded genital
band, are placed in 3-4 layers, the one above the other, and have an irregular roundish
polyhedric shape, and measure 6, I/O, 5' in diameter. Each single follicle (fig. 40) is
surrounded and separated from the others by a thin fulcra! lamella containing nuclei (zs).
In transverse section, under stronger magnifying power, we see that the larger mother-
cells of the spermatozoa (sd), which arise from the endodermal epithelium of the sinus
wall, line the wall of the follicle, whilst the centre is filled by the ripe spermatozoa (PI.
XXIII. fig. 40 sz).

Periphema} Hseckel, 1877.

Peripkyllidae, with four perradial buccal pouches of the oesophagus and four perradial
niches of the basal stomach, united in its aboral basis. Between the niches, the four
subumbral funnel cavities (or the four hollow interradial tseniola of the basal stomach)
form hollow cones, which are beset with two rows of gastral filaments, but are free from
them above each end, separated below the point of the cone.

I established the genus Periphema in 1877 (in the Prodrornus Systematic
Medusarum) for a large Periphyllid, of which there were, unfortunately, only broken and
incomplete fragments of a single, very large specimen in the Challenger collection. I
was, however, able by careful examination of these fragments, and with the help of other
Periphyllidse examined by me (viz., by comparison with the large, perfectly preserved
specimen of Periphylla mirabilis), to compose a complete quadrant of the Medusa from
the fragments, from which the figure in Plate XXIV. is drawn in its natural size. The
reconstruction was more difficult, as the enormously developed proboscis or buccal stomach
(fig. 3) was completely torn away from the pylorus and broken in pieces, and there
were also distracting abnormal deformities — clearly in consequence of an earlier but
completely healed injury — on the only remaining quadrant of the subumbrella (fig. 1),
which I have of course left out in the figure. Apart from these, our Periphylla regina
seems very closely allied to the preceding Periphylla mirabilis, and I therefore included
it without hesitation in this genus in my System (1879, p. 421). However, I now
consider it more appropriate to separate it generically from Peripiliylla regina under the

1 Yltqttpnfios = greatly renowned.

REPORT ON THE DEEP-SEA MEDUSAE. 85

uanie of Periphema regina, as I had already done in the Prodronius (1877). Whilst in
the true Periphylla (P. mirabilis, P. hyacinthina, &c.) the four large interradial
conical funnel cavities of the subumbrella traverse the whole length of the central
and basal stomach, and meet with their points in the centre of the umbrella cone,
in Periphema regina they stop short a little way below the subumbrella, so that the
four points of the conical funnel cavities remain separated by a basal hollow space, which
has the geometrical basis of a quadrate pyramid, and fills the point of the cone. The four
perradial niches of the basal stomach are consecpaently connected at their aboral ends
by this conical cavity, whilst they are completely separated in the true Periphylla.

Periphema regina, Hseckel (Pis. XXIV., XXV.).

Periphema regina, Hreckel, 1877, Prodrom. System. Medus., No. 389.
Periphylla regina, Hfeckel, 1879, System der Medusen, p. 421, No. 423.

Umbrella bell-shaped, nearly as high as broad. Pedal zone of the exumbrella rather
narrower than the lobe zone, both together nearly as high as the cone zone. Marginal lobes,
oval, rounded obtusely, their distal wings nearly semicircular, about half as high as their
proximal gelatinous swelling. The eight tentacle lobes project further on the umbrella
margin than the eight rhopalia lobes. Tentacles very thick, nearly as long as the height
of the umbrella, one-third as broad at their base as the marginal lobes. (Esophagus
cubical, very large and very thick-walled, nearly half as high and half as broad as the
umbrella, the oral margin in the plane of the umbrella margin, without barbous filaments.
Horizontal diameter, 180-200 mm. ; vertical diameter, 180-200 mm.

Habitat. — The Antarctic Ocean, south-west of the Kerguelen Islands. Lat. 62° 26' S.,
long. 95° 44' E. Station 156. The large specimen, to which the fragments examined
belonged, was a mature female, and was taken from a depth of 1975 fathoms, 26th
February 1874. The colour of the broken fragments, otherwise well preserved in
spirit, was reddish, the ovaries were brownish-yellow and the endodermal epithelium
of the abaxial wall of the coronal sinus — or the inner surface of gelatinous umbrella —
from dark red-brown to black-brown.

The umbrella (PL XXIV. figs. 1, 2) of Periphema regina, as far as could be made
out from the fragments to hand, is bell-shaped, considerably more depressed than in
Periphylla mirabilis. Its apex is flatly truncated, and nearly ecpial in height to the
diameter of the bell opening, 18-20 cm. The exumbrella is divided by a broad, deep
coronal furrow (fig. 2, ec), nearly in the middle of the height, into an upper umbrella
cone, and a lower umbrella corona. The umbrella cone is smooth, flattened above, and
almost hemispheroidal. The coronal furrow is very broad, and the gelatinous substance
of it very much thinned. It is divided by sixteen subradial longitudinal furrows (which
pass below into the lobe clasps) into sixteen broad, crescentic arese ("areola) semilunares,"
fig. 2, ec). A vane-like-shaped process of the exumbral zonal muscle (mz) lies between

86 THE VOYAGE OF H.M.S. CHALLENGER.

the crescentic folds at the end of each longitudinal furrow. The umbrella corona may
be subdivided into an upper pedal girdle and a lower lobe girdle. The pedal girdle
("zona pedalis") is, however, much less strongly developed than in most other Periphyllidse,
and hardly 2 cm. high. The pedalia are slightly vaulted and comparatively small. The
lobe girdle of the umbrella corona (figs. 1,2," zona lobaris ") is more strongly developed
in Periphylla regina and differently shaped from that of Periphylla mirabilis. The sixteen
subradial marginal lobes are much larger compared to the pedalia, and are rounded, not
pointed. The difference in size between the four pair of ocular lobes, and the four
pair of tentacle lobes alternating with them is considerably greater. Neither the two
gelatinous swellings lying in each marginal lobe nor the inter-furrow are so thick as in
Periphylla mirabilis; the lobe clasp (" loboporpa," fig. 2, kl), which lies at the bottom
of this furrow, and supports the septum between the two halves of the pouch, is much
feebler, but shows the same structure in transverse section, fig. 10, comp. above, p.
71), on the other hand, the thin delicate wings (" patagia," Ip), which form the selvage
of the lobe margin, are much broader and longer in our species than in the foregoing.
If we measure from the circular line of the exumbrella, indicated by the insertion of the
tentacles between the marginal lobes, the tentacle lobes are 50 mm. long (without wings,
35 mm.), and the ocular lobes only 45 mm. (without wings, 30 mm). The tentacle lobes
are more than 30 mm. broad in the middle, the ocular lobes a little over 20 mm. (figs.
1,2).

The four interradial sense clubs in this species appear to be very small and almost
rudimentary (fig. 2, o) ; in the fragment, to hand, however, there was only one preserved,
and it did not allow of closer investigation. Only half of the twelve tentacles were
preserved (figs. 1, 4). They are on the whole of the same nature as in the preceding-
species (comp. above, p. 67), but are considerably shorter and thicker. Their length is
nearly equal to the height of the umbrella (18-20 cm.), whilst in Periphylla mirabilis it
is twice as great. The longitudinal muscle appears to be less strongly developed. The
thickness of the hollow tentacles at the conically swollen base amounts to 10 mm. They
then thin away into a cone and run out below into a fine point (fig. 2, t). The
peculiar insertion of the tentacle by two root muscles (mk) inside a tentacle funnel (it), and
the remarkable formation of the double-valved vent-hole at its base, is the same here as
has been already described in Periphylla mirabilis (comp. PI. XXII. fig. 22, and PI.
XXV. fig. 1).

The inner concave umbrella wall (subumbrella, PI. XXIV. fig. 1 ) in Periphema regina
shows on the whole the same conditions already (p. 71) described in detail in Periphylla
in irabilis. The muscular system is, however, much more strongly developed in the former,
and the separate muscles show more prominently. The eight longitudinal deltoid muscles
("musculi deltoidei") are yellowish-white, very stout, firm glistening bands, and appear re-
markably powerful. The strongest is the interradial deltoid muscle (fig. 1, md'), an equila-

REPORT ON THE DEEP-SEA MEDUSAE. 87

teral triangle, 36 mm. in height, 32 mm. at base, whose truncated point reaches as far as
the middle of the genitalia, and is there inserted at the interradial septal node (hi). The
muscular fibres which diverge radially from its point towards the base are nearly equally
powerful throughout. The narrow " musculus intergenitalis " (ms) above, between the two
genitalia of each pair, is formed by a weaker process of this deltoid muscle. The perradial
deltoid muscle (md') is weaker than the interradial; it forms an equilateral triangle 20 mm.
in height, 25 mm. at base, whose truncated point reaches as far as the oral end of the
gastral opening (go), and is inserted there below the palatine groove (gs) at the perradial
palatine node (gk). The lateral muscular fibres (md'") are much more strongly developed
in this muscle than in the median muscles. A band-shaped "musculus congenitalis" (mp)
springs from each side of the perradial deltoid muscle ; it lies coradially between the outer
margin of each genitalium (s) and the gastral opening (go), and extends to the upper end
of the latter. This band-shaped congenital muscle is 1 0 mm. broad below, 5 mm. broad
above ; its length amounts to 60 mm. Its fibres which run parallel, and only converge
slightly above, spring from the lateral margin of the perradial deltoid muscles, and are
inserted above at the pyloric opening (gy).

The broad coronal muscle (fig. 1, mc) shows essentially the same condition as that
already described in Periphylla mirabilis (p. 71). Its proximal margin (mc) serves
as a basis of origin for the deltoid muscles. Its subumbral surface is elevated into
10 to 12 circular folds (mc2) with deep furrows sunk between them (mc3). The coronal
muscle is also divided by the sixteen subradial lobe clasps into sixteen coronal arese.
These are 25 mm. high in the middle (between each two marginal lobes), but 30 mm.
high laterally in the middle of each marginal lobe. The four ocular muscular area?
(25 mm. broad) are only a little smaller than the twelve tentacular coronal area?
(30 mm. broad). The intermediate lobe clasps are much weaker than in the preceding
species, but show the same structure in transverse section (fibrous cartilage, PI. XXV.
figs. 9, 10). The formation of the lower or distal margin of the coronal muscle (PI.
XXIV. fig. 1, wm;4) is peculiarly differentiated. Whilst in Periphylla mirabilis it is
quite smooth, projects internally like an umbrella roof over the insertion of the tentacles,
and forms a simple tentacle funnel (it), in Periphema regina it is fringed and divided
into numerous fine folded lobes or " frenula." On each of the sixteen muscular arese
there are nearly twenty such frenula, 2-3 mm. long, which connect the distal margin of
the muscle with the subumbral surface of the marginal lobe lying below it. An equal
number of subumbral funnel-shaped depressions are deeply inserted between these
frenula in the thickened distal margin (" infundibula subcoronaria ").

Both parts of the umbrella cavity, the lower simple coronal umbrella cavity and
the upper quadrilocular funnel umbrella cavity, comport themselves the same on the
whole in Periphema regina as in Periphylla mirabilis. The simple coronal umbrella
cavity forms a circular hollow space, whose subumbral external wall forms the umbrella

88 THE VOYAGE OF H.M.S. CHALLENGER.

corona with the coronal muscle and the deltoid muscle lying above it. The wide hollow
space of the coronal umbrella cavity is filled, for the most part, by the powerful buccal
stomach, whose oral margin extends to its opening. The quadrilocular funnel umbrella
cavity (which is sharply defined by the four palatine nodes (gk) from the simple umbrella
coronal cavity) shows however an essential variation in our species. Whilst in Periphijlla
mirahilis the four conical interradial umbrella funnels traverse the whole length of the
central and the basal stomach, and meet above in the central point of the umbrella cone,
in Pervph&ma regina they stop some little way below the cone; the four points of the
funnels are here inserted separately at four interradial points of the umbral wall of the
flattened basal stomach, which are at 4 cm. distance from each other. This occasions a
perfectly different formation of the basal stomach, which chiefly justifies the foundation
of the genus Periphema.

Apart from the differentiated formation of the basal stomach, the gastrovascular
system of Periphema regina shows essentially the conditions already described in detail
in Peripkylla mirabilis. Only the special formation of single parts and their comparative
sizes show unimportant differentiations. Of the three chief sections of the axial principal
intestine, the buccal stomach is the largest, being 8 mm. high, whilst the height of the
central stomach and of the basal stomach only amounts to 5 cm.

The buccal stomach or oesophagus (fig. 3) is extremely fleshy and thick walled. The
four quadrants of the oesophagus were found as four isolated fragments, still partially
connected with the pieces broken off from the central stomach, in the bottle containing the
incomplete remains of our species. One such quadrant is represented in natural size in
fig. 3. Each quadrant contains a complete buccal pouch {bb), and the enclosing half of
the oral column touching it (ee). From the beast having been torn during its capture,
the oesophagus was quartered through the interradial meridian planes. The reconstructed
form of the buccal stomach is on the whole that of a cube of 7 cm. to the side ; more
closely considered, it forms rather an octagonal prism with alternating broad and narrow
lateral surfaces ; the former are formed by the buccal pouches, the latter by the oral
columns. The oral columns (" columnar buccales," fig. 3, ac), are remarkably strong, and
supported by a powerful, fleshy, gelatinous swelling. The adradial wings of the oral
columns (" alee buccales," ad) appear extremely fleshy, and laid internally in strong
longitudinal folds, whilst their interradial middle plate is thinner, very much extended
and diminished in size, towards the oral margin. The wings project internally con-
siderably above the lateral parts of the buccal pouches, so that they are arched out
on both sides into spacious wing pouches. The buccal pouches (" bursse buccales,"
bb), when inflated would be almost hemispheroidal : their wall is supported by a thin,
elastic but firm, gelatinous plate, which is broadened below and rounded obtusely
at the oral margin. The distal ends of the eight adradial wings therefore project
most below at the oral margin (am), without running out into barbous filaments as in

REPORT ON THE DEEP-SEA MEDUSAE. 89

the preceding species. The unusual strength of this large proboscis indicates a predatory
mode of life.

The central stomach appears capable of being completely shut off from the buccal
stomach, as the palatine opening {gp) is narrowed by strongly-projecting palatine
swellings, and both the four perradial palatine nodes (gk) and the contiguous lateral parts
of the palatine grooves are also considerably thickened. On the other hand, the four
obelisk plates of the central stomach are very delicate and thin walled (torn for the most
part). The four perradial angles of the quadrate pyloric opening coincide with the
four proximal ends of the four cleft-shaped gastral openings.

The basal stomach (gb) shows an essentially different formation from that of the pre-
ceding species. In the latter the four perradial peripheric niches surrounding its
conical axial space are completely separate from each other, whilst the four interradial
funnel cavities of the subumbrella run above as far as the point of the conical basal
stomach and meet there in the centre of the umbrella cone. In Periphema regina, on
the other hand, the interradial funnel cavities end 2 cm. below the basal centre point of
the basal stomach. The latter consequently forms a quadratic undivided depression in
the bottom of the flatter vaulting of the cone with the funnel cavities and their phacelli
springing from its four angles. The distance of these four points (the lateral length of
the quadrant) amounts to 4 cm. The shattered condition of the fragments before me,
did not allow of the complete reconstruction of the basal stomach. The central part of
the umbrella cone with its four funnel points was, however, preserved, and showed clearly
that the four perradial niches of the basal stomach communicate freely. This
peculiarity distinguishes Periphema generically from Periphylla.

The phacelli or longitudinal rows of gastral filaments in Periphema regina are
extremely large and more strongly developed than in any other Medusa known to me.
They consist of several thousand strong and very long filaments, placed in several rows
along the gastral tseniola (not in a single row as in the preceding species). The filaments
are longest in the middle of the phacelli, up to 80 mm. long and 1 mm. thick. They
become shorter and thinner towards both ends, and are then mostly only 10-20 mm. long
and hardly 0"5-0-2 mm. thick. Their special formation and distribution is the same as
in the previous species. Two diverging phacelli run from the cone point of each of the
four interradial funnels, extending on the lateral margins of the gastral openings as far
as the pylorus, and ending 1 cm. above the palatine groove (fig. 1). The filaments are
sometimes cylindrical, sometimes flattened like a ribbon, often thickened into knots and
tongue-shaped at the end. The nature and disposition of these glands is the same as in
the preceding species. The lumen both of the central and the basal stomach is occupied
for the most part by this mass of filaments.

The peripheric coronal intestine in Periphema regina shows the same formation as
that already described in detad in Periphylla mirabilis (p. 78). The colossal coronal

(ZOOL. CHALL. EXP. PART XII. 1881.) M 12

90 THE VOYAGE OF H.M.S. CHALLENGER.

sinus (cs) which only communicates with the central stomach by the four perradial
gastral openings, is divided a little above the middle into four quadrants by the four
inter radial septal nodes (fig. 1, hi). These " cathammal nodes " are only a few millimeters
large, but consist of very firm fibrous cartilage (cornp. above, pp. G7, 80 ; and PI. XXV.
fig. 8). The peripheric pouch corona, into which the coronal sinus opens at its lower
margin by sixteen tranverse clefts (at the upper margin of the coronal muscle), is divided
by the sixteen subradial lobe clasps into sixteen coronal pouches ; and each of these is
subdivided by the invagination of the tentacle funnel into an inner and an outer coronal
pouch (axial velar pouch and abaxial avelar pouch). Besides these, each coronal pouch
gives out two lobe pouches below, which compose the marginal " festoon canal " ; and
whilst each of the four interradial coronal pouches sends an ocular pouch to the sense
club, each of the twelve remaining coronal pouches sends out a wide tentacle canal into
each tentacle (comp. above, p. 81, and the explanation of PI. XXV. fig. 1).

Genitalia (PI. XXIV. fig. 1, sf). The fragment before me belonged to a mature
female, but only one pair of the four pairs of reproductive glands was preserved. The
two ovaries of this pair showed the situation and form represented in the middle of fig. 1.
They lay between the gastral openings in the subumbral wall of the coronal sinus, whose
upper and lower margin they almost touch with both ends. Both ovaries of the pair Lie
almost parallel beside each other in the upper half, and are only separated by the narrow
intergenital muscle (5 mm. broad). On the other hand they diverge strongly in the lower
half, as there the triangular interradial deltoid muscle (md/) is inserted between them.
The distance between the lower ends amounts to 50 mm. Each of the eight ovaries
forms a narrow horseshoe-shaped arched genital band, whose convex distal arch nearly
touches the upper margin of the coronal muscle (mc') below, whdst the two parallel
limbs, which lie close together, almost reach above to the pylorus (gy). The thickened
supporting plate of the subumbrella forms a projecting midrib (" sterigma, costa
genitalis," st) in the middle between the two limbs. The genital band is raised on both
sides into a series of folds, which project internally into the urnbrella cavity and
externally into the coronal sinus (figs. 5, 6). The number of these broad folds, which are
subdivided like a fan into smaller folds (figs. 5, 6), amounts from 40 to 50 in each ovary
(20 to 25 in each limb). They are 4-6 mm. long, 2-4 mm. broad, and closely packed
with spheroidal ova. The smallest ova lie at the basal margin of insertion of the folds,
the largest at the freely projecting margin, which is turned towards the " costa genitabs "
(st). At the basis of the folds we see clearly that the smallest and youngest ova
originate immediately from the encloderm cells which line the subumbral wall of the coronal
sinus. As soon as the ova grow to a certain size, each ovum becomes enclosed in a
gelatinous fulcral sheath (fig. 7, yz), a superficial abaxial growth of the supporting plate
of the subumbrella (»•:). In transverse sections, through the genital folds, we see the
ova, enclosed in these fulcral capsules, lying in rows beside one another (fig. 7). The

REPORT ON THE DEEP-SEA MEDUSAE. 91

extension and strength of the fulcra] capsule increases proportionately with that of the
ovum enclosed. The ripest ovum of Periphema regina reaches the extraordinary size
of a millimeter and more. The ova consist for the most part of an opaque food yolk,
composed of spheroidal yolk granules of equal size (O'Ol mm. diameter), thickly com-
pacted (fig. 4, yd). Each ripe ovum is also enclosed (inside the fulcral sheath) by a
thick structureless (?) chorion (fig. 4, yc), showing a projecting micropyle at one spot
(fig. 4, ym). It has the form of a short bottle neck, and resembles the micropyle known
in the eggs of our freshwater mussels (Naiadacea). Below the micropyle we can
distinguish with the naked eye a white spot (" cicatricula ") on the yellow yolk, in which
the large spheroidal germinal vesicle is enclosed ("nucleus," yn). It contains a visibly
dark, germinal spot (" nucleolus," yf), and this again contains a large double contoured
germinal point (" nucleolinus," fig. 4, yp).

Order VIII. CUBOMEDUS^E, Haackel, 1877.

Acraspedse with four perradial sense clubs, containing an auditory club with
endodermal otolite sac and one or more eyes ; four interradial tentacles or bunches of
tentacles. Stomach with four wide perradial quadrangular pouches separated by four
long, narrow interradial septa or fused selvages. Genitalia four pair of leaf-shaped swell-
ings, which are fastened by one margin along the four interradial septa, are developed
from the subumbral endoderm of the gastral pouches, and project freely into their hollow
space.

Family Charybdeid^e, Gegenbaur, 1856.

ChArybdeid.e, Hasckel, System der Medusen, 1879, p. 433, taf. xxv.

Cubomedusas with four simple, interradial tentacles and four perradial sense clubs ;
without marginal lobes in the velarium, but with eight adradial marginal pouches ; with-
out pouch arms in the four broad perradial pouches.

Sub-family, Tamoyid^:, Hseckel, 1877.
Charybdeidaa with velar canals, and with four perradial frenula of the velarium.

Charybdea,1 Peion and Lesueur, 1809.

Charybdeidas with four simple interradial tentacles, having pedalia ; with suspended
velarium (with velar canal and four perradial frenula). Stomach flat and low, without

1 X«pi;/35is = an eddy, a gulf, rapacious.

92 THE VOYAGE OF H.M.S. CHALLENGER.

broad mesenteries ; central stomach and basal stomach fused, without distinct pyloric
stricture. Four horizontal groups of filaments, simple or double, bush-shaped or brush-
shaped, limited to the interradial corners of the bottom of the stomach.

The genus Charybdea, the oldest known genus of this family and order, was
founded by Peron in 1809, with the following indefinite diagnosis : — "La concavity de
l'estomac se confondant avec celle de l'ombrelle ; rebord garni de faux bras, ou
plutot de faux tentacles" (Tableau des Meduses, &c, Annal. Mus. H. N., vol. xxiv.
p. 332). Peron united in this genus two entirely different Acraspeda, both of which
he knew only very superficially and incompletely — the Mediterranean Charybdea
marsupialis and the secmatorial Atlantic Charybdea periphylla. The latter was
first seperated by Steenstrup and raised to be the representative of the genus
Periphylla. On the other hand, the genus Charybdea was retained by almost all
new authors for the known Charybdea marsupalis of the Mediterranean, which had
already been described and figured by Plancus in 1739, as " urtica soluta marsupium
referens," and of which Milne-Edwards had given a very full (though for the most part
mistaken) description in 1833. Quite recently (1879) Claus gave a very detailed
histological monograph of this type of the genus Charybdea. I was myself able
to examine several new species of this genus, and to re-describe its character more
minutely. In the sense which I have retained here, those Charybdeidae which have
a suspended velarium (with canals and frenula) belong to the Charybdea. Charybdea
is distinguished from the genus most nearly related {Tamoya) by the flat, low pouch-
shaped stomach, the narrow mesenteric folds, and, specially, by the formation of the
gastral filaments. These are distributed horizontally in the four perradial corners of
the bottom of the stomach, as four simple or double pencil-shaped or brush-shaped
groups of filaments, whilst in Tamoya they extend as four vertical bands in the inter-
radial lateral lines of the large depending gastral sac. The deep-sea species described
below is, on the whole, nearly related to the Mediterranean Charybdea marsupialis,
which is only half the size, but is distinguished from it by the broader velum, containing
twice as many velar canals, which are also much more richly dendritic. Moreover, the
sculpture of the exumbreUa is different. The histological conditions have been
described in great detail by Claus in his monograph on Charybdea marsupialis; we shall
therefore confine ourselves to a short account of the organological peculiarities, giving
special prominence to the specific differences shown between Charybdea murrayana and
Charybdea marsupialis. There may perhaps be sufficient to justify this species being
taken as the representative of a separate genus : Charybdusa. I have named this
species in honour of my friend John Murray, first assistant in the Challenger
Commission.

REPORT ON THE DEEP-SEA MEDUSAE. 93

Charybdea murrayana, Hseckel (PI. XXVI.).

Charybdea murrayana, Hseckel, 1879, System der Medusen, p. 442, No. 436.
CJiarybdusa murrayana, Hseckel, 1877, Prodrom. System Medus., No. 408.

Umbrella bell-shaped, almost cubical, rather higher than broad, depressed above,
somewhat widened below ; lateral surfaces almost quadratic. Stomach quite flat with
four short oral lobes, four tuft-shaped phacelli, composed of bunches of large brush-
shaped filaments. The vertical distance of the heart-shaphed sense niches from the
umbrella margin half as great as the horizontal distance of the pedal bases. Velarium,
broad with twelve dendritic velar canals in each quadrant. Pedalia, a longish oval, one-
third as long as the height of the umbrella. Tentacles cylindrical, longer than the height
of the umbrella. Horizontal diameter of the umbrella, 50 mm. ; vertical diameter,
60 mm.

Habitat. — West Coast of Africa, not far from Sierra Leone. Lat. 30° 10' N., long.
14° 51' W. Depth, 200 fathoms. Station 348. There were two well-preserved female
specimens, taken 9th April 1876.

The umbrella (PI. XXVI. figs. 1, 6) is, on the whole, nearly cubical, as in most Cubo-
medusee. The vertical diameter (60 mm.) is however rather greater than the largest
horizontal diameter (50 mm.) ; four rounded, interradial " corner pillars" (corresponding
to the tentacles) project more or less strongly at the four vertical lateral corners of the
cube, whilst the four perradial lateral walls (corresponding to the pouches) recede between
the pillars and seem more depressed. As they do not lie quite vertically but diverge a
little below, the umbrella has really the shape of a truncated, regular quadrilateral
pyramid. Its upper apical surface is slightly depressed and circumscribed bke a cap,
by a horizontal coronal furrow.

The exumbrella, as in most Cuboniedusee, is divided by longitudinal furrows into a
number of arese, projecting convexly between the furrows. We can generally distinguish
sixteen such exumbral furrows, viz., firstly, eight subradial furrows (fig. 1, ea), which
separate the four broader perradial lateral walls from the four narrower corner pillars ;
secondly, four perradial furrows, which divide the four depressed lateral walls in two and
extend downwards from the ocular crypt to the velarium (fig. 1, ep) ; and thirdly, four
interradial furrows, which halve the four projecting corner pillars (ei). The last-named
furrows are the deepest, so that the two halves of each pillar project in the form of
semi-cylindrical swellings. The four pedalia, which bear the tentacles, run out below
from the pillars. The four perradial sense clubs, alternate regularly with the pedalia,
and lie high above the umbrella margin in a special cavity of the exumbrella, the sense
niche (" crypta rhopalaris," eo). The external heart-shaped entrance to this deeply
hollowed crypt is partly covered by the scale of the rhopalium (" squama rhopalaris "), a
protective scale of the exumbrella, projecting above the opening like a roof. The
exumbrella appears finely granulated, as numerous urticating warts or round groups of

94 THE VOYAGE OF H.M.S. CHALLENGER.

thread cells are scattered freely over it. The gelatinous substance of the umbrella shows
a considerable degree of firmness, in spite of its being very thin and without any cellular
elements. The gelatinous substance varies in thickness in different places, according to
the different longitudinal furrows of the exumbrella and the subumbrella, being thinnest
along the interradial furrows (in the middle of the corner pillars) and thickest at the two
sides of the pillars, and above in the cap-shaped apical cover of the umbrella (figs. 1-3, ag).

The subumbrella or nectocalyx is nearly cubical. The four corners of the cube are
interradial and formed by the narrow septa of the broad gastral pouches, or by the
"fused streaks" by which the subumbrella is connected with the umbrella. The
muscular layer of the subumbrella is thus divided into four rectangular muscular plates,
which are placed nearly vertically to each other in the interradial " fused streaks "; they
correspond to the four lateral surfaces of the cube, and form the axial wall of the four
radial pouches (fig. 3, mw). The circular fibres of each muscular plate are, however,
interrupted in its perradial middle line by a band-shaped, longitudinal muscle, which
extends from the ocular niche, upwards to the mesogonia and downwards to the
frenulum (fig. 3, mp). The broad coronal muscle is therefore actually divided here into
eight quadrangular coronal areae as in Pericolpa (System, taf. xxiii.). Whilst, however
in Pericolpa, these areae lie in the principal radii (four perradial and four interradial), in
Charybdea they are placed adradially.

The umbrella margin (figs. 1,5, 8), in a wider sense, bears four perradial sense clubs
and four interradial tentacles. These marginal organs are connected by a remarkable
nerve ring of peculiar structure. Below this nerve ring, however, the umbrella margin
passes into a broad velarium, a thin marginal membrane resembling the velum of the
Craspedotae, but, however, essentially different. As regards the eight marginal organs
they are undeniably derived phylogenetically from the eight principal tentacles of
Tessera and Tesserantha (PI. XV.) ; the four sense-clubs from the four perradial principal
tentacles, and the four tentacles from the four interradial. In this respect the condition
is exactly inverted in the Cubornedusae, as in the Peromedusae (specially in the
Pericolpidae). In the Discomedusae all the eight principal tentacles are transformed into
rhopalia.

The velarium or marginal membrane (figs. 2, 5, 8, va), represents a membranous,
annular distal process of the umbrella margin. It has hitherto been simply termed
velum, and placed beside the similarly termed velum of the Craspedotae. These two
formations are, however, only analogous, not homologous ; they have originated independ-
ently of each other, and their structure though similar is in no way identical, that is,
their relation to the nerve ring is essentially different. As in all Charybdea belonging to
the sub-family of the Tarnoyidae, the velarium is traversed by special canals, and is
fastened in a very peculiar fashion to the subumbrella by the four perradial frenula
(suspensors or supporting folds, figs. 2, 8, vf). These frenula are muscular, vertical,

REPORT ON THE DEEP-SEA MEDUSAE. 95

gelatinous folds, formed by a visible, perradial thickening of the gelatinous supporting
plate, and stretching from the sense depression to the free margin of the velum. They
keep the velarium suspended horizontally, and can raise it still higher by contraction of
their longitudinal muscles. The velarium is divided by the four perradial frenula on the
one hand and the four interradial pedalia on the other, into eight adradial octants or
" velar lobes." These are homologous in position and morphological importance, with
the eight free marginal lobes of the Pericol'pa, and the eight arms of the LucernaridEe
(comp. Lucernaria, Pis. XVIL, XVII., and also my System, taf. xxii., xxiii.). Hence
we see that the velarium of the Cubomedusas corresponds to a corona of eight fused
adradial marginal lobes.

The umbrella cavity (figs. 2-6) is almost cubical, corresponding to the subumbrella.
Its four vertical sides are formed by the subumbral walls of the four radial pouches, the
upper surface by the subumbral gastral walls ; the lower surface is occupied by the
umbrella-opening, which is strongly contracted by the projecting velarium. The
stomach hangs down iu the axial space of the umbrella cavity ; its peripheric space is
divided above into four small interradial funnel cavities (" infundibula"). These are
formed in the upper (proximal) part of the umbrella cavity in such a way that they
stretch across the four perradial mesogonia (which we shall describe below) from the
four corners of the stomach to the middle of the four radial pouches. The frenula of the
velarium correspond to these proximal suspensors in the lower distal part ; four corre-
sponding niches are sunk as velar funnels between the frenula. The horizontal
diameter of the umbrella disk is consecmently smallest in the four centripetal projecting
perradial lines, largest in the centrifugal projecting interradial lines (along the cathammal
septa) ; the former correspond to the lateral lines of the quadrate, the latter to the
diagonal lines.

The pedalia, or gelatinous sockels (figs. 1-5, tvi), are four peculiarly-shaped inter-
radial gelatinous appendages of the umbrella margin. They bear the tentacles at the
distal end, and are sharply defined from them. Gegenbaur terms the sockels of the
Charybdea " marginal leaves," Fritz Muller " processes of the corner swellings," and Claus
" umbrella lobes." Claus compares them erroneously with the marginal lobes of the
other Acraspeda. But these true marginal lobes never He in the principal radia of
the first and second order (perradial and interradial), but always between them. On
the other hand, the peculiar pedalia of the Cubomedusae always he interradially, and can
only be compared to the pedalia in the Peromedusas, which bear both tentacles and sense
clubs (comp. above, p. 65). In our Cliarybdea murrayana (figs. 1-5, tvi) the pedalia
are cuneiform or trilaterally prismatic in the upper third, compressed laterally in the two
lower thirds, and shaped like a thin longish oval leaf, nearly a third as long as the height
of the umbrella ; its axial edge is curved concavely, its abaxial edge convexly, whilst its
lateral surfaces appear bent unsymmetrically. The tentacle springs from its truncated

9(5 THE VOYAGE OF H.M.S. CHALLENGER.

distal end ; the thicker proximal end is cut out concavely, and inserted at the lower part of
the corner, swelling above the umbrella margin in such a manner that a small axial
cavity, or pedal funnel (" infundibulum pedale," fig. 3, it), remains between the two.

The four tentacles are strong, cylindrical, hollow filaments, 4 mm. thick, thickened
like a club at the basis (to 6 mm.), and longer than the height of the umbrella (probably
several times as long in the uninjured animal). In the longitudinal and transverse
sections, their thick wall shows the same peculiar and complicated structure, fully
described by Claus in Charybdea marsupialis.

The four perradial sense clubs or marginal bodies (" rhopalia ") lie above the umbrella
margin, in the ectodermal sense niches (" crypta rhopalaria or ocularia") already men-
tioned. The structure of these highly-developed organs of sense in Charybdea murrayana
is the same as in the Mediterranean Charybdea marsupialis, where they were first inves-
tigated by Gegenbaur in 1836, and recently and minutely by Claus in 1878. They have
a very complex structure, and essentially resemble those of the Peromedusae and Discome-
dusse, as they contain both optical and acoustic organs ; their finer structure, however,
varies in several respects, and in some ways very peculiar. Each sense club is fastened
by a thin peduncle into the sense niche of the exumbrella, and is partly covered
externally by the protective scale, which projects like a roof over the exodermal aperture
of the rhopalar niche. It contains a large otolite sac containing numerous crystalline
endodermal otolites in its club-shaped swollen terminal part. The six eyes, two larger
unpaired in the perradial middle line, and four smaller paired on the two sides of the
unpaired, lie above the otolite sac ; each unpaired eye consists of a pigment cup, a thick
lens, and a powerful corpus vitreum lying between them ; the lens is wanting in the
smaller paired eyes. A very large ganglion opticum of a highly developed structure
forms the nerve centre of the optical apparatus.

The nervous system has the same high centralisation as in the other CubornedusEe, and
corresponding to their highly developed organs of sense, it shows itself in a more com-
plete and more centralised form, than in the other Acraspeda ; in this respect it attains
the highest stage of formation among all Acraspeda. The central nervous system, which
was discovered in Tamoya by Fritz Muller (1859), consists of a complete nerve ring and
of eight ganglia, the four larger perradial being placed at the basis of the sense clubs,
and the four interradial at the basis of the tentacle pedalia ; from the perradial ganglia
sense nerves go out to the organs of sense and motor nerves to the longitudinal muscles,
while motor nerves go out to the tentacles from the interradial ganglia. The former
always lie considerably higher than the latter, so that the nerve ring rises in a vaulted
arch from the rhopalar niche to the basis of the pedalia. The whole nerve ring (figs. 2-8,
re) therefore forms four depressed arches. Their highest part lies perradially, their
lowest part interradially. The nerve ring Hes embedded in a groove of the subumbrella,
interrupting its muscular plate, and consists of a clear axial cord and two more turbid

REPORT ON THE DEEP-SEA MEDUSAE. 97

fibrillar cords (an upper and an under) lying on it with the peculiar nerve epithelium lying
above them. Extensive plexuses of fibrillse with large multipolar and spindle-shaped
ganglion cells run out thence and spread chiefly on the subumbrella. The finer
structure of the nervous system and the organs of sense have been recently described in
detail by Claus in Charybdea marsupialis (1879, loc. cit.). His endeavour to compare the
condition of this structure of the Cubomedusae with that of the Craspedota, is, however,
untenable, as the two have arisen independently of one another, and are, therefore, not
homologous. The nerve ring of the Cubomedusae also corresponds only to the lower
(suburethral) nerve ring of the Craspedotse, whilst the upper (exumbral) ring of the former
is entirely wanting. On the other hand, the central nervous system of the Peromedusae
is probably essentially closely allied to that of the Cubomedusse.

The gastrovascular system (figs. 1-10) resembles that of the Stauromedusse in
the simplicity of its formation (Tesserantha, PI. XV. ; Lucernaria, Pis. XVI., XVII. ). The
principal stomach or axial intestine is connected by four horizontal perradial gastral
openings with four wide quadrangular radial pouches, which are divided in their entire
length by four narrow interradial septal selvages, and communicate by a narrow circular
canal at the distal end of the selvages. The axial principal intestine, or the stomach in
the wider sense (" gaster principalis "), really consists in most Cubomedusae of the same
three sections as in the Stauromedusee and Peromedusaa, viz. , an aboral basal stomach, a
middle central stomach, and an oral buccal stomach ; the pyloric opening (" pylorus/'
gy) also forms in this case the boundary between the basal and the central stomach
and the palatine opening (" palatum," gp), that between the central and the buccal
stomach. In Charybdea, however, as in many other Charybdeidae, the pyloric open-
ing is very wide and the pyloric stricture very slightly developed, so that, taken together,
the basal and the central stomach seem to form a single, simple, somewhat flat, quadratic
chamber.

The buccal stomach or oesophagus (" gaster buccalis," go) — the " oral funnel " of Fritz
Muller, " oral peduncle " of Claus — is comparatively small in our species, and forms a flat
quadrate pyramid. Its truncated point is formed by the narrow palatine opening (fig. 9,
gp), its angles by the four perradial strong oral ribs, thickened selvages of the gelatinous
plate, which gives consistence to the whole stomach. The oral ends of these buccal ribs
project considerably at the quadrate oral opening, and cause the formation of the four
lanceolate or oval " oral lobes." A deep perradial groove runs on the axial endodermal
surface of these frilled triangular oral lobes ; it bends with a sharp turn towards the out-
side at the palatine opening, and runs, enclosed in the mesogonial fold, on the inner sur-
face of the subumbral wall of the central stomach as far as the middle line of the radial
pouch (figs. 4, 6, gs). The thickened oral rib itself, which at the same time forms the
midrib of the leaf-shaped many-folded oral lobe, runs at the palate immediately into the
low mesogonial fold. The folded oral tubes, which were strongly contracted in our

(ZOOL. CHALL. EXP. PART XII. 1881.) M 13

98 THE VOYAGE OF H.M.S. CHALLENGER.

spirit specimen and appear thickly frilled at the margins, are probably capable of greater
expansion in the living animal.

The central stomach in this Charybdea, as in most Charybdeidse, is joined to the
basal stomach, as the pyloric stricture between the two is not developed and only faintly
indicated by the slightly projecting pyloric valves. These two divisions of the stomach
therefore compose a wide, but very flat pouch, or a low chamber, quadratic in outline.
Its bottom or lower wall represents the thin quadrate plate, which at the same time forms
the fundus of the cubic umbrella cavity. This muscular plate is pierced in the middle of
the palatine opening, from whose four perradial corners the gastral grooves already men-
tioned (figs. 4, 6, gs), run to the middle of the four gastral openings. The horizontal cover
of the low gastral chamber or its upper wall is formed by the smooth endodermal surface
of the cap-shaped umbrella apex (figs. 2, 3, git). The four interradial corners are occupied
by the four pyloric valves, the narrow " bow-shaped fused lines " (Glaus), which are placed
perpendicularly at the proximal ends of the long septal selvages. On the other hand, the
four perradial side walls of the chamber between the selvages are represented by the four
gastral openings (fig. 6, go), four narrow horizontal clefts, which lead from the stomach
into the four radial pouches. We find here a complicated arrangement of valves, by
means of which the stomach can be completely shut off for a time from the radial pouches,
These four perradial " pouch- valves " alternate with the interradial pyloric valves (gy).
Above each pyloric valve the stomach forms a peculiar evagination in the form of a low
triangular pouch, and the phacelli or dendritic bunches of gastral filaments (b) are placed
in this pyloric pouch.

The gastral filaments (fig. 7,f), are much more strongly developed in Charybdea
■murrayana than in the closely-allied Charybdea marsupialis; in each of the four inter-
radial corners i if the stomach they form a visible phacellus or bush, composed of ten to
twelve larger and several smaller branches. The stems of these branches are connected
below at the root, where they rise from the aboral surface of the subumbral pyloric valve,
and so actually represent the principal branches of a single, very short, powerful stem, a
primary interradial primitive filament. The lower (distal) half of each branch consists of
a strong, simple, or bifurcate stem, the upper (proximal) half of a pencil-shaped bunch of
numerous branches, which are partly simple, partly dichotomised (figs. 9, 10). The solid
axis of the filaments is formed by a thick cylindrical or flat ribbon-like gelatinous filament
(a process of the supporting plate of the subumbrella) ; its endodermal epithelium is mostly
composed of gland cells, having many flagellate cells at the base and urtricating cells at
the point.

The four broad quadrangular radial pouches (figs. 2-6, bp), occupy the greater
part of the subumbrella, and are only separated from each other by four narrow inter-
radial septal selvages (Jcs). These correspond to the septal nodes of the Tesseridae and
Peromedusae, and to the septal selvages of the Lucernaridse ; and, like the latter, have

REPORT ON THE DEEP-SEA MEDUSAE. 99

arisen by fusion of the umbral and suburnbral wall of the primitive stomach of the
Scyphostoina, (" cathamma," h). The remains of the gastral epithelium are therefore
visible in the transverse section of the selvages, in the form of the " enclodermal lamella,
cathammal plate, or vascular plate" (fig. 10, hp), which separates the thicker gelatinous
disc of the umbrella (ug), from the thinner supporting lamella of the subuinbrella (zw).
We can even distinctly distinguish two layers of cells in the gastral plate, of which the outer
belongs to the umbral endoderm, the inner to the suburnbral endoderm. A leaf-shaped
genitalium, which projects freely into the contiguous radial pouch, is fastened along the
entire length of the septal selvages on each side of its suburnbral part (fig. 10, s).

Four margins and two walls can be distinguished in each radial pouch. Whilst the
two lateral margins of the quadrangular pouch are formed by the interradial septal
selvages, its lower (or distal) margin is the proximal velar margin and its upper (or
proximal) margin is the gastral opening. The latter can be completely closed by the
perradial pouch-valve ; this is formed by a horizontal fold of the subumbrella, which
rises at the upper margin of the pouch and projects as a thickened gelatinous plate freely
into the cavity of the basal stomach. The external or abaxial wall of the radial pouches
is formed by the smooth enclodermal surface of the gelatinous umbrella, its inner or axial
wall by the delicate subumbrella. The latter is thin-walled and very extensible, and
consists from within to without of the usual four layers: — (1) The enclodermal
epithelium with high, glandular, cylindrical cells (fig. 10, dw) ; (2) the thin but firm sup-
porting plate or gelatinous lamella (zw) ; (3) the muscular plate (mw) ; and (4) the
exodermal epithelium (qw). Although pretty firm, the suburnbral wall is so thin that it
stretches, like a delicate vefl, above the pouches, and allows all the organs lying in them
to shine clearly through. A narrow band-shaped longitudinal muscle (fig. 3, mp) runs
in its perradial middle line. This muscle passes above into the " mesogonium," or upper
supensory, below into the " frenulum velarii," or lower supensory. The latter divides the
distal section of each of the radial pouches into two broad adradial lobe pouches.

The eight lobe pouches or marginal pouches (" bursse lobares," or " marginales," figs.
2, 3 8, bin) are caused by a perradial septum, which, running from each rhopalar niche
to the upper velar margin, divides the distal part of each radial pouch into two halves.
This septum is merely the abaxial margin of the frenula itself in wThich the umbral and
suburnbral walls of the pouch are fused together. Each of the marginal pouches thus
formed is rectangular, nearly twice as broad as high. Dendritic, csecal, velar canals run
from their lower or distal margin into the "velarium" (fig. 8, cv). These lie entirely
in the thickened supporting lamella of the velarium, and are flattened like a ribbon ;
their endoclermal epithelium, like that of the radial pouches, is flat and clear on the
umbral side, high and glandular on the suburnbral. Their ramification is delicately
dendritic and is weaker towards the perradius, stronger towards the interradius. There
are forty-eight velar canals on the whole, so that twelve of them come on each

100 THE VOYAGE OF H.M.S. CHALLENGER.

quadrant. The largest velar canal lies nearest the interradial pedal and shows 6 to 8 pairs
of side branches, partly simple, partly cleft. On the other hand, the number and size
of the irregular side branches increases at intervals towards the frenulum (fig. 8). The
velar canals lie freely in the gelatinous fulcra! lamella of the velarium and are not
connected by a cathammal plate ; they are therefore secondary formations, which have
subsequently grown with the solid supporting plate of the velarium from the distal
margin of the lobe pouches.

The four perradial rhopalar canals (or ocular vessels) arise by a funnel-shaped basis
from the middle line of the radial pouches above the velar frenula, and pass, narrowed,
immediately into the peduncle of the sense club, in whose free head part they end in
an ampulla-shaped expansion. The four tentacle canals (or the pedal canals leading
into the tentacles) arise at the four interradial angles of the umbrella by a double
root, as each tentacle receives a root canal from the distal corner of each quadrangular
radial pouch. Each pouch therefore gives out two root canals for two adjacent
tentacles. The junction of the two root canals takes place immediately below the
distal end of the septum. The tentacle canal proceeding from it traverses the entire
length of tentacle, and is comparatively very narrow, owing to the thickness of the
tentacle wall. A kind of marginal circular canal is formed by the communication of
the radial pouches, which is produced at their distal margin by the root canals.

The genitalia (PI. XXVI. figs. 2, 6, 10, s) form eight broad, thin, semi-oval leaves
which are fastened in pairs along the four interradial septal selvages, and project freely
from these into the four radial pouches ; they occupy the greater part of their hollow
space so that the two reproductive leaves of each pouch touch each other or even overlap
with their free margins in its middle (fig. 2, s). Claus sees in this formation "a
very peculiar arrangement" (1879, Zoologie, p. 289). The difference presented
between the reproductive glands of the Cubomedusse, and those of the other Acraspeda,
is, in fact, only insignificant ; and the former may easily be referred back to the latter.
Most Lucernaridse show the same conditions in the broader anatomical sense, as in these
two genitalia come upon each of the four broad radial pouches. These, however, do not
belong to the said pouches, but rather to the interradial septum, which separates each two
pouches. The two genitalia, which belong to two adjacent pouches and are separated
by a septum, form one pair, and in Halicyathus, as in Tesserantha (PL XV.), are con-
nected into a horse-shoe by a convex arch at the proximal end of the septum. In the
remaining Lucemaridse (Pis. XVI., XVII.) this U-shaped connective arch has undergone
retrograde formation, so that eight separate adradial reproductive leaves lie beside each
other, and this holds good for the Peromedusae and Cubomedusae. In all cases, likewise,
the reproductive elements are formed from the subumbral endoderm of the radial
pouches (fig. 10, rw). Then the reproductive leaves are fastened to the septum in
such a way that they touch the umbral wall immediately ; but as they are completely

REPORT ON THE DEEP-SEA MEDUSA. 101

separated from it by the cathammal plate of the septum (ks), they belong genetically to
the subumbral wall. Each reproductive leaf is really a thin fold of this suburethral wall,
inasmuch as the gelatinous supporting lamella of the latter forms a leaf-shaped process
(" sterigma," fig. 10, zs), which is covered on both sides by the subumbral endodermal
epithelium. The reproductive elements, which fall freely when ripe into the pouch, are
developed from the subumbral endodermal epithelium on both sides of the sterigma.
They originate from the subepithelial endoderm cells, which cover both surfaces of the
federal process. The latter corresponds to the sterigma or fulcral frame in the genitalia
of the Peroniedusse (p. 83), and formed in both female specimens before me a broad,
fibrous, axial plate (fig. 10, zs), thickly covered with later and earlier egg cells (so). The
reproductive leaves are covered on both free surfaces by the connected cylindrical
epithelium of the endoderm (fig. 10, pd). The ripe reproductive elements pass from
the radial pouches into the stomachs through the gastral openings, and are expelled
through the mouth.

o

Order VIII. DISCOMEDUSiE, Hseckel, 1866.

Acraspeda with eight to sixteen or more sense clubs (always four perradial and four
interradial, besides these occasionally several accessory clubs) ; in each sense club an
auditory club with an endodermal otolite sac and often an eye at the same time.
Marginal lobes always eight pair of primary (Ephyra lobes) and frequently numerous
accessory (velar lobes) besides. Tentacles sometimes present, sometimes wanting.
Stomach surrounded by a corona of radial processes (8 to 16 to 32 or more; sometimes broad
radial pouches, sometimes narrow radial canals. Genitalia four interradial folded
swellings in the subumbral gastral wall, developed from its endoderm (rarely divided
into eight adradial swellings) ; sometimes invaginated in the form of a pouch towards the
inside in the central gastral cavity, sometimes evaginated hernia-like towards the outside
in the umbrella cavity. Umbrella depressed and discoid. The general fundamental form
of all Disconiedusse is the octomeral Ephyra.

Fikst Sub-order of the Discomedus^:, CANNOSTOM^E, Hasckel, 1879.

Discomedusse with undivided proboscis or oral tube, a simple, quadrangularly prismatic
oesophagus, without oral arms ; with simple or quadrangular central oral opening, and
with short, solid marginal tentacles.

Family, Ephyeid^;, Hseckel, 1877.
Ephtridj:, Hreckel, System der Medusen, 1879, taf. xvii., xviii., p. 450.
Cannostomse with broad radial pouches, without terminal branched canals. Discome-

102 THE VOYAGE OF H.M.S. CHALLENGER.

dusae with simple, quadrangular oesophagus, without oral arms; with simple oral opening,
usually sixteen broad radial pouches (eight ocular and eight tentacular), more rarely
32 to 04. Usually eight sense clubs (four perradial and four interradial), more rarely 16
to 32. Alternating with these an equal number of short, solid tentacles. Usually 16
(rarely 32 to 64) marginal lobes, with or without simple lobe pouches, always without
branched lobe canals; four interradial or eight adradial genitalia in the subumbral gastral
wall.

Sub-family, Nausithoid^e, Haeckel, 1879.

Ephyridae with eight sense clubs and eight adradial tentacles, with sixteen marginal
lobes and eight separate adradial genitalia.

Nauphanta,1 Haeckel, 1879.

Ephyrid with eight sense clubs and eight tentacles, with sixteen marginal lobes and
thirty-two lobe pouches (sixteen ocular and sixteen tentacular). Central stomach
opened by four perradial gastral openings into a ring sinus, from whose distal margin
run out sixteen coronal pouches ; eight separate adradial genitalia, equally distributed,
not grouped in pairs.

The genus Nauphanta is, as yet, represented only by the remarkable deep-sea
Medusa described below. It is most closely allied to the Mediterranean Nausithoe
among all Medusas hitherto known, but is distinguished from it by peculiar conditions
of structure. The sculpture of the exumbrella with its deeply insected coronal furrow
between the central disc and the peripheric corona, and with the very prominent pedalia
(polyhedric gelatinous swellings between the radial furrows) reminds us strikingly of the
Periphyllidse and Collaspidae ; in other respects it appears to be a very old intermediate
form connecting among Peromedusae, Cubomedusae, and Discomedusae ; as it is closely
related morphologically to all these groups, it indicates the common descent of the
Ephyroniae and Tesseroniae. Nauphanta takes the highest place among the three genera
of the Nausithoidae, and represents the most highly developed form among the octomeral
Ephyridse. In many respects it approaches the following polymeral Collaspidae. It
agrees with the closely-allied Zonephyra and Pelayia in having thirty-two lobe pouches,
whilst it differs from them both in the formation of the reproductive organs. These
comport themselves the same as in Nausithoe, and form eight roundish adradial sacs,
similar in form and at equal distance from each other. The two specimens before me, a
male and a female, are perfectly mature. The ovaries are eight tuberous, scutiform plates,
whose endodermal upper surface is covered with very large ova. Instead of these plates
the spermaria form numerous digitate spermatic sacs. The developed pedalia of the
corona of the exumbrella remind us of the Peromedusae (Periphylla) on the one hand,

1 NatxpivTTi, the name of a ship in Aristophanes.

K.EPOKT ON THE DEEP-SEA MEDUSAE. 103

and of the Collaspida (AtoJla) on the other. Like Atolla, Nauphanta is a true deep-sea
form of high phylogenetic antiquity.

Nauplianta challengeri, Hseckel (Pis. XXVI. , XXVIL).

Nauphanta challengeri, Hreckel, 1879, System der Medusen, p. 487, No. 452.

Umbrella cap-shaped, with a horizontal apical surface, and vertical side- wall, one and a
half times as broad as high. Exumbrella with a deep coronal furrow and sixteen deep radial
furrows. Umbrella corona with sixteen pedalia (eight smaller rhopalar and eight stronger
tentacular) ; sixteen marginal lobes oval, nearly twice as long as broad, with a deeper
peronial furrow, about one-fourth as long as the radius of the umbrella. Tentacles
cylindrical, pointed, about as long as the radius of the umbrella. Genitalia eight oval,
adradial, kidney-shaped swellings, twice as long as broad ; their proximal halves some-
what broader than their intervals, their distal halves covered by the coronal muscle.
Horizontal diameter, 12 mm. ; vertical diameter, 8 mm.

Habitat. — The South Atlantic Ocean, not far from the island of Tristan da Cunha.
Lat. 32° 24' S., long. 13° 5' W. Depth, 1425 fathoms. Station 335. Both specimens
examined (a male and a female) are well preserved, and were taken on 16th March
1876. The transverse and longitudinal sections figured are taken from the two halves
of the halved female specimen.

The umbrella (PI. XXVIL fig. 1 ; PI. XXVIII. figs. 12-14) of Nauphanta
challengeri has the form of a cap or biretta, and is considerably more vaulted than in
most other Discomedusas. Its special conformation, and especially the peculiar
sculpture of the exumbrella, reminds us in many respects of the Cubornedusas and
Peromedusas, with which the oldest ancestral forms of the Discomedusas are clearly very
closely allied. Whilst the upper flattened apical surface appears almost horizontal,
the steep, vertical side walls stand almost vertical. The umbrella is constricted
between the first or second third of its height by a deep horizontal coronal furrow
(" fossa coronalis," ec), and is thereby divided into an upper (central) umbrella disc,
and a lower (peripheric) umbrella corona. The umbrella disc (" discus umbrellas "),
which is depressed above like a cup in the middle, forms the horizontal cover of the
flat discoid basal stomach (gb) ; the umbrella corona (" corona umbrellas ") encloses the
corona of the radial pouches, and bears below at the margin the corona of the tentacles,
and rhopalia, and the marginal lobes alternating with them.

The exumbrella (figs. 1, 13) is distinguished by the horizontal coronal furrow (ec),
and also by deep, radial, or longitudinal furrows, which, as raPeriphylla (Pis. XIX., XX.),
divide the external surface of the umbrella into convex, projecting, gelatinous swellings.
We can distinguish on the whole sixteen deep, subradial, longitudinal furrows, and
sixteen shallower, alternating with them. The latter traverse nearly the whole

104 THE VOYAGE OF H.M.S. CHALLENGER.

exumbrella and touch both the umbrella disc and the umbrella corona, whilst the deeper
subradial furrows are confined to the umbrella corona. Of the sixteen shallow, longi-
tudinal furrows, four are perradial, four interradial, and eight adradial. They are placed
at equal distances in the central umbrella disc, and divide its peripheric, thickened half
into sixteen equal, subradial disc-swellings, whilst its thinner central half remains
without furrows, and is, at the same time, considerably thinned away (figs. 1, 13, 14).
In the peripheric umbrella corona, on the other hand, they are only distinctly impressed
in the distal part of the sixteen coronal swellings. The deeper sixteen subradial
longitudinal furrows, which traverse the entire corona of the umbrella, lie between the
sixteen coronal swellings, and are placed in pairs in such a way that the umbrella
corona is divided into eight narrower and eight broader gelatinous sockels or pedalia ;
the former bear the eight rhopalia, the latter the eight tentacles (figs. 1, 13). Each of
the sixteen gelatinous sockels consists of a thicker, undivided proximal part and a
thinner distal part, halved by a shallow, radial furrow ; the former contains a coronal
pouch, the latter a pair of lobe pouches. The eight narrower, principal ocular sockels
(" pedalia rhopalaria "), four perradial {up), and four interradial {wi), are distinguished by
their side lines being sinuated concavely, and their narrower proximal part being only half
as long as the bifurcate distal part. The eight broader adradial tentacular sockels
(" pedalia tentacularia," ww), on the contrary, show convexly projecting side lines, and
their broader proximal part is nearly twice as long as the deeply inserted distal part
(figs. 1, 13). The ends of the bifurcate halves are rounded obtusely in all sixteen
pedalia, and sharply defined from the marginal selvage of the marginal lobes
(" patagium," Ip).

The subumbrella (figs. 12, 13) is divided into three sections by the broad coronal
muscle (one), by whose two margins they are separated from one another. Its inner or
upper intracoronal third reaches from the insertion of the stomach (relatively from the
four pyloric valves or interradial septal nodes) as far as the inner or proximal margin
of the coronal muscle (mc), and contains both the proximal halves of the eight adradial
genitalia (s), and the narrow longitudinal deltoid muscles alternating with them ; of
these muscles, as in Atolla (PL XXIX.), the four interradial (fig. 12, md) are much
stronger and broader than the four perradial (fig. 12, md). The middle or coronal third of
the subumbrella is occupied solely by the broad coronal muscle ("museums coronalis," figs.
12, 14, mc). This comports itself precisely as in Periphylla, and is divided by the
sixteen fused clasps of the marginal lobes (fig. 12, hi) into sixteen quadrangular coronal
area?. Of these the eight adradial (tentacular) are considerably broader than the eight
principal (rhopalar) ; the former cover, at the same time, the distal halves of the
genitalia on their axial side. The external or lower extracoronal third of the
subumbrella extends from the outer or distal margin of the coronal muscle (me4) to
the actual margin of the umbrella, and is occupied by the corona of lobes. In it

REPORT ON THE DEEP-SEA MEDUSAE. 105

we can distinguish sixteen pairs of longitudinal lobe muscles, a pair for each marginal
lobe.

From the umbrella being so much vaulted, the umbrella cavity (w) is more spacious
and higher than in most other Discomedusae. It is nearly cylindrical in form, as its
subumbral side walls rise nearly perpendicular (fig. 14). But as the eight genitalia
project like arches towards the inside, it is rather octangularly prismatic. Its upper
base is occupied by the subumbral bottom of the stomach (gw), its lower bases by the
wide opening of the umbrella, surrounded by the corona or marginal lobes. The axial
middle space of the proximal half is filled by the pendant oesophagus. The
subumbral gastral wall forms four narrow mesenteric folds or mesogonia in a radial
direction above, and projects further between them iD an interradial direction, so as to
form four flat, interradial funnel cavities (fig. 3, ii) ; these are covered over by the four
flat pyloric valves (fig. 2, gi), which bear the phacellse (fig. 2, f). The special
formation of this part is very similar to that of many Cubomedusse (Charybdeidae).

In Nauphanta, as in Ephyra, the common ancestral form of all Discomedusse, and in
most genera of the family Ephyridas (all Palephyridse and Nausithoidea), the umbrella
margin (PI. XXVII. fig. 1; PL XXVIII. figs. 12-14) is regularly composed of the
following marginal organs : — Eight rhopalia (four perradial and four interradial), eight
adradial tentacles alternating with these, and sixteen subradial marginal lobes, inserted
between the rhopalia and the tentacles. The number of the sixteen marginal organs,
which alternate with the sixteen subradial marginal lobes, is therefore the same here as
in Tesseranbha (PI. XV.) and Periphylla (PI. XVIII. &c). Whilst, however, in the
Stauromedusa Tesserantha all the sixteen marginal organs remain simple tentacles, and
in the Peromedusa Periphylla the four interradial tentacles are transformed into
rhopalia, in our Ephyrida only the eight adradial tentacles appear to be permanent ;
the eight principal tentacles (four perradial and four interradial) are transformed into
the characteristic sense clubs, as in all other Disconiedusse.

The eight sense clubs or rhopalia (figs. 12, 13, c?',-fig. 20) resemble most strongly
those of the most closely allied Nausiihoe, among all known forms of these organs,
though they also agree in many and most important points with those of Periphylla
(PL XVIIL). They are distinguished from those of most other Discomedusae by their
broad, succinct shape. The eight sense clubs lie hidden between each pair of marginal
lobes in four perradial and four interradial deep incisions of the umbrella margin, which
alternate with the deep tentacular incisions (figs. 12-14). Each rhopalium has, on the
whole, the form of a broad tongue-shaped leaf, and is nearly one and a half times as long
as broad. In the normal position of the vertical umbrella margin, its free distal end is
directed upwards in such a way that the convex abaxial surface looks freely outwards,
the concave axial surface freely inwards towards the umbrella cavity. Of the four sense
organs which are united in each rhopalium, the olfactory depression lies on the convex

(ZOOL. CIIALL. EXP. PART XII. 1881.) M 14

106 THE VOYAGE OF H.M.S. CHALLENGER.

external side of the basal part, the eye opposite on its concave internal side, the tactile
plate below the eye, and the free auditory club hidden in the spacious auditory niche
(fig. 20, on). The olfactory depression or olfactory funnel (" infundibulum olfactorium,"
oz) forms a flat conical depression in the convex exumbral side of the thickened basal
part ; its endodermal epithelium is laid in delicate folds, and consists of rod-shaped sense
cells (olfactory cells?). Opposite it, on the concave subumbral side, there is a broad black
brown pigmented pad (fig. 20, op), in whose centre the unpaired axial eye lies embedded,
as in Na usithoe ; this seems to contain a concave-convex lens in the middle of a darker
pigmented knob (oc). Below the knob a narrow dark pigmented band runs out,
which projects more strongly convexly, bears a variously shaped epithelium with long
tactile hairs, and probably represents a tactfle plate (op). The auditory club (oh) rises on
a thin stalk outside this plate (on its abaxial side) ; it hangs freely down in the concave
rhopalar niche (on), and is surrounded protectively towards the outside by the broad
concavo-convex protective scale or auditory scale (os) ; the blunt lower margin of the
latter is folded over above towards the inside. The solid auditory club, whose ectodermal
epithelium bears long auditory hairs, encloses a spheroidal or subspheroidal otolite
(fig. 20, ol ; fig. 21) in its free swollen distal end. This otolite is crystalline, and trans-
parent and shows many irregular, polygonal, slightly convex facets, as well as a
sharply projecting granulation on its upper surface. Several smaller otolites seem added
to the larger one at the proximal end.

The eight tentacles (t), which alternate with the eight rhopalia, and therefore He
adradially, spring further above in deeper incisions of the umbrella margin. They are
nearly as long as the height of the umbrella, cylindrical, pointed like an awl at the
distal end, and swollen to a cone at the proximal basis. A short canal (a branch of the
eight adradial coronal pouches) runs some way into the basal part of the tentacles which
otherwise are solid. Their principal mass forms a soft, elastic, chordal axis, composed of
large vesicular endoderm cells. The ectodermal covering consists partly of thread cells,
partly of tactile cells, and partly of epithelial muscular cells. The long muscular fibres of
the latter run longitudinally and form a strong longitudinal muscle on the axial side of
the tentacles.

The sixteen marginal lobes (hn) lie subradially, in the middle between the eight adra-
dial tentacles and the eight alternating rhopalia. They are obliquely oval, with unecpial
sides, as their tentacular margin is nearly twice as long as their rhopalar margin. Each
marginal lobe is considerably thickened in its proximal half, by the inverted bifurcate
branches of each two adjacent pedalia, whilst its distal half is formed by a very delicate,
thin-membraned, almost triangular patagium (lp>).

The gastrovascular system (PI. XXVII. figs. 2-10; PL XXVIII. figs. 12-15) of
Nauphanta appears at first sight very simply formed, and not essentially different from
that of Ephyrula, the known common germinal form of the Discomedusse. On closer

REPORT ON THE DEEP-SEA MEDUSA. 107

investigation, however, it shows several very remarkable and important conditions of
formation not to be found at the present day in the majority of Discomedusse, and which
may be considered extremely old peculiarities inherited from the common ancestral form
of the Acraspedse. In this way Natiphanta comes nearer the Tesseronise than the other
Ephyroniae, and connects these two sub-sections of the Acraspedae in a most interesting
fashion ; above all, it is remarkable in one respect, that the four important interradial
septal nodes or cathamma which separate the four broad perradial gastral pouches and
which have disappeared entirely in most Discomedusse, still exist here. The reproductive
glands lie in the subumbral wall of the coronal intestine below the septal nodes (much
further, therefore, towards the exterior than in most other Discomedusge) But in the
peculiar nature of the central principal intestine, and also in that of the peripheric coronal
intestine, we find manifold peculiarities which recall the Tesseroniaa more than the
Ephyronise, and which must be regarded as very ancient heirlooms from the common
ancestral form of the two sections.

The axial principal intestine (" gaster principalis," figs. 2-7) appears at first sight to
consist, as in the other Discomedusse, of two principal sections, of the upper (aboral)
central stomach and the lower (oral) buccal stomach ; the former is covered by the
umbrella disk, and is itself flatly discoid ; the latter is more funnel-shaped, and hangs
freely down in the umbrella cavity. The buccal stomach is, however, constricted in the
middle ; this stricture probably corresponds to the palatine opening (" porta palatina,"
gp), in which case we can probably still distinguish here all the three gastral chambers
of this section of Medusa?. The boundary between the two principal sections is formed
by the horizontal cathammal plane, in which the four septal nodes or cathamma (Jen) are
placed ; these may be considered the pyloric opening (" porta pylorica," gy). Other-
wise the three gastral chambers have an extremely simple formation. If the foregoing
supposition be correct, the buccal stomach or oesophagus (" proboscis ") is limited to the
oral half lying below the palatine opening (gp), which has the form of a truncated
quadrangular pyramid. The base of the latter is formed by the quadrate oral opening,
from whose four corners the four perradial short triangular oral lobes project (figs. 12, 14,
al). It only extends as far as the proximal margin of the coronal muscle ; consequently
the oesophagus only occupies the upper half of the umbrella cavity. Above the palatine
opening (gp), the stomach is again dilated in the form of a flat funnel, corresponding to
the true central stomach (gc). This funnel opens above immediately into the flat basal
stomach (gb), and appears only separated from it by the four interradial pyloric valves
(" vavulae pylorica?," gi). These are four flat tongue-like projections, which stand out
centripetally from the four septal nodes in the base of the stomach, and bear the gastral
filaments at their upper free end (fig. 14;</) ; they completely correspond to the stronger
pyloric valves of many Cubomedusa? (p. 98). The ideal horizontal plane, in which
they lie, corresponds to the pylorus of the Tesseroniae, and therefore actually forms the

108 THE VOYAGE OF H.M.S. CHALLENGER.

lower boundary surface of the basal stomach, whose upper surface is formed by the
horizontal, almost level, endodermal surface of the central gelatinous disc of the umbrella
(fig. 14, ng).

The four septal nodes ("nodi cathammales," fig. 3, kn; fig. 14, kn) are four
interradial, small but firm nodules, hard as cartilage, in which the suburnbral gastral
wall is firmly fused with the umbral. Four broad horizontal clefts remain between
the nodes, the four perradial gastral openings, by which the central stomach
communicates with the peripheric coronal intestine. These important conditions of
organisation correspond clearly to those of the Tesseroniae. The four important
interradial cathammal nodes especially, as well as the ring sinus lying beneath them,
are homologous with those of the Peromedusae ; whilst on the other hand the condition
of the four interradial tongue-shaped pyloric valves shows a special homology with
many Cubomedusas. In most other Disconiedusse (certainly in all Semostomce and
Rhizostomce) these Tesseronia-like formations have disappeared, as the septal nodes
and the pyloric valves have undergone retrograde formation.

The gastral filaments (PI. XXVIII. fig. 18) are not very numerous, but com-
paratively large and thick. They are four arched interradial phacellae, whose convex
margin corresponds to the free margin of the tongue-shaped pyloric valve (gi). Each
crescentic phacella consists of a single row of from twenty to twenty -four gastral
filaments, placed closely near each other. They are cylindrical, and decrease in length
from the middle of the phacella towards the two ends, the longest one nearly one-
third as long as the radius of the umbrella.

The peripheric coronal intestine (" gaster coronaris ") extends from the horizontal
cathammal plane of the four septal nodes (which lies a little above the exumbral
coronal furrow), to the umbrella margin and consists of the following three coronal or
horizontal sections : A, a proximal or upper corona of four perradial gastral pouches ;
B, a middle corona of sixteen coronal pouches ; and C, a distant or lower corona of
thirty-two marginal lobe pouches. On closer examination we can even distinguish
five different sections of the coronal intestine, as there is also a special coronal sinus
(cs) between the four perradial principal pouches and the sixteen coronal pouches, and
below it an intercalary corona of eight pouches (fig. 4). Compare the transverse section
(figs. 2-10) and the longitudinal section (figs. 14, 15).

The four broad radial pouches, or perradial gastral pouches (fig. 4), which belong
to all Tesseronia, are undeniably also present here, or are at least represented by the
four gastral openings or broad cleft spaces between the four interradial septal nodes
(kn). We may also include with these the circular hollow space below the septal nodes,
between them and the proximal margin of the coronal pouches, unless we prefer to
compare this hollow space to a special coronal sinus (cs, fig. 14), Hke that of
the Peromedusaj ("sinus coronaris," p. 79, PL XXL, cs). In fact the conditions here

REPORT ON THE DEEP-SEA MEDUSAE. 109

are the same as in all those Tesseronise, whose iuterradial septa or cathammata are merely
small short nodes, not long selvages.

The sixteen coronal pouches (" bursas coronares ") which form the middle zone of

the coronal intestine, go out from the distal margin of the coronal sinus; they are

alternately broader and narrower, and are divided from each other by the sixteen

subradial septal selvages, which form the proximal processes of the sixteen subradial

lobe clasps (figs. 4-15, M). Their inner or axial wall forms below the sixteen coronal

areas of the coronal muscle, its outer or abaxial wall the sixteen pedalia of the

umbrella. Each coronal pouch divides below (at the distal margin of the coronal

muscle) into three csecal terminal branches, of which the two lateral enter the

inverted halves of the two adjacent marginal lobes, whilst the middle passes either

into a rhopalium or into a tentacle. The eight narrower ocular pouches (bo) (four

perradial and four iuterradial) extend to the eight rhopalia, in whose ampullae their

middle terminal branch ends cascally. The eight broader tentacle pouches (bt) are

much wider, and contain the distal halves of the genitalia, which are fastened to their

subunibral wall ; their middle terminal branch passes into the basal part of the

tentacles ; they project a considerable way into the umbrella cavity (figs. 5,6). Of

the eight ocular pouches, the four perradial are somewhat longer than the four

iuterradial, as they spring from the coronal sinus rather higher up than the latter.

This explains how in the transverse section of the umbrella only eight radial pouches

appear immediately below the simple coronal sinus (fig. 4). The four narrow,

perradial ocular pouches (bo^ alternate with four very broad wide pouches (br2) which

contain the upper proximal ends of a pair of genitalia ; rather further down they

divide into three pouches, a middle iuterradial ocular pouch and two lateral adradial

tentacular pouches (fig. 5). This comportment of the coronal pouches is best seen

by comparative consideration of the longitudinal section (figs. 14, 15) and the transverse

section (figs. 3-8). It is also worthy of remark that the subunibral endoderm of the

ocular pouches rises into high papillae and folds, corresponding to the sterigma of the

tentacular pouches (figs. 4-8).

The thirty -two lobe pouches (" bursas lobares," hi) fill in pairs the proximal half
of the sixteen marginal lobes, whilst the delicate thin-membraned distal half of the
lobes remains free (figs. 12, 14, hi). The two pouches of each lobe are separated by the
subradial lobe clasp (hi) and belong to two different adjacent coronal pouches, an Oakl-
and a tentacular. As the rhopalia lie considerably deeper than the insertions of the
tentacles, the sixteen ocular lobe pouches are much shorter than the sixteen tentacular
lobe pouches. The distal ends of both lie, however, in the same horizontal plane. The
peripheric corona of pouches in Nauphanta, therefore, shows essentially the same
conditions of formation as in Pelagia.

The genitalia (PI. XXVII. figs. 4-8, s; PI. XXVIII. figs. 12-16, s) in both sexes

110 THE VOYAGE OF H.M.S. CHALLENGER.

form eight separate bean-shaped glands, lying regularly distributed internally on the
subumbral wall of the coronal intestine, above the eight tentacles. More minute
investigation, however, shows that they are associated in pairs, as in the Cubomedusse
and Peromeduste. Consequently, there are really four interradial pairs of genitalia
present, which originally stood in immediate relation to the four septal nodes. We see
clearly, especially from transverse sections, through the proximal halves of the ovaries, a
little above the coronal muscle, that the eight genitalia really form four interradial pairs
which have been developed from the four interradial septal nodes. Each pair of genitalia
lies in a broad interradial pouch (br'), where the four interradial ocular pouches are still
united with their two tentacular pouches (fig. 4) ; and rather further down, the sterigrua
of the two associated genitalia are curved and rolled inwards in such a way that their
convex, lobed upper surfaces are turned towards one another. The two reproductive
glands of each pair consequently correspond to arched halves of the four interradial
genitalia of Tesserantha. The form both of the ovaries and the spermaria in
Nauphanta is bean-shaped or kidney-shaped, concave on the axial side, convex on the
abaxial. They extend above into the coronal sinus, near the septal nodes with the upper-
most parts of their truncated proximal half, whdst they almost touch the distal margin
of the coronal muscle and the tentacle basis, with the lowermost part of their thinner
distal half. The two halves are separated externally from each other by the proximal
margin of the coronal muscle (figs. 12, 14, mc), which stretches like a veil above the
lower balf. At a superficial view, it seems as if the genitalia lay in the subumbral wall
of the coronal intestine, and from thence form projecting pouches in the umbrella cavity.
Comparison of longitudinal and transverse sections shows, however, that for the most part,
they lie freely in the hollow space of the tentacular coronal pouches and are only con-
nected with their subumbral wall at a node-like point, which we shall call the genital root
(figs. 4-11, 15, stx). The fulcral frame (" sterigma "), bearing the endodermal germinal
epithelium, runs out at this root from the gelatinous supporting lamella of the subumbrella.
The sterigma (st) or the fulcral frame of the genitalia runs out from the root as
a short, thick cone ; it immediately extends in the shape of a thin, strong, arched
shield, having many folds, and bearing numerous irregularly-formed hollow papillae on
its convex upper surfaee. This fulcral frame of the genitalium then appears branched
dcndritically both in the transverse and the longitudinal sections (4-11, fig. 15), it
corresponds to the pinnated genital rib of the Peromedusse ( " sterigma," p. 83, PI.
XXIII. fig. 38). The node-like root of the sterigma is crescentic, cut out concavely at
the upper or proximal margin. At the same time, it is hollowed out by a csecal arching
outwards of the coronal pouch, in such a way that in the transverse section (fig. 6) it
seems to begin with two separate radical branches, which are the two horns of the
crescent (st"). The shape of the sterigma is, therefore, really very complicated
(figs. 2-15). The cartilaginous connective tissue, which forms the fibrous stroma of the

REPORT ON THE DEEP-SEA MEDUS/E. Ill

sterigma, has numerous cells, especially towards the endodermal upper surface. As the
sterigma is only connected with the subumbrella at its thin root, it otherwise projects
freely into the hollow space of the coronal intestine, extending into the coronal sinus
with the proximal half, and into the eight tentacular coronal pouches with the distal
half. The sterigma is covered on its convex outer surface by the usual endoderm of
the subumbral wall of the coronal intestine. On the concave inner surface, which
encloses a genital sinus with repeated archings-out, this endoderm is transformed into
germinal epithelium, which forms reproductive elements.

The germinal epithelium of the endoderm (fig. 16, ds), which forms large egg cells
in the female, and spermatic follicles nearly twice the size in the male, is found
exclusively on the concave inner surface of the shell-shaped, bent-in fulcral shield, as
lining of the genital sinus enclosed by the shield. The sinus has a very complicated
form, as the scutiform sterigma is turned over concavely , not only at its abaxial (outer),
but also at its lateral surfaces, so that it is repeatedly arched outwards. The folded-
over, concave, axial half is, moreover, fused for the most part with the convex abaxial
half in such a manner that only a narrow passage leads from the hollow space of the
coronal intestine into that of the genital sinus. This narrow passage is the " apertura
sinus genitalis " (sa) ; it is difficult to find, and appears both in Nauphanta and in
Atolla to lie turned towards the two genitalia belonging to an interradial pair, at the
interradial side of the sterigma! root. This aperture was the more difficult to find in
the two preserved spirit specimens, as the hollow space of the sinus was almost entirely
filled with coagulated slime (?), and the epithelium in great part destroyed. Both the
ripe egg cells and the ripe balls of spermatozoa pass from the germinal epithelium into the
gelatinous plate of the sterigma, and are enclosed here in thin-walled fulcral capsules.
These burst later on, and the ripe reproductive elements probably do the same, then
fall directly into the hollow space of the coronal intestine of the genital sinus, through
whose aperture they are emptied into the coronal intestine, from thence by the gastral
openings into the stomach and then outside through the mouth.

Sub-family, Collaspid^e, Haeckel, 1879.

Ephyridas, with 16 to 18 sense clubs, and the same number of tentacles, with 32 to
64 marginal lobes, and 8 separate genitalia.

Atolla,1 Hseckel, 1879.

Ephyrida, with 16 to 32 rudimentary sense clubs, and the same number of
tentacles, with 32 to 64 marginal lobes, and 64 to 128 lobe pouches. Central stomach
opened by four perradial gastral openings into a coronal sinus, from whose distal

1 Atolla — an island surrounded by coral reefs.

112 THE VOYAGE OF H.M.S. CHALLENGER.

margin run out 16 to 32 broad tentacular coronal pouches, and the same number of
alternating rudimentary ocular canals ; 8 separate adradial genitalia, grouped in pairs,
not distributed at equal distances.

The genus Atolla, like the preceding Nauphanta, is one of the most remarkable
and morphologically interesting deep-sea Medusa brought to light by the Challenger
expedition. Both are very ancient remains of an extinct ancestral group of Discomedusae,
which clearly indicate the close connection of this order with the Cubomedusae and
Peromedusas. Atolla has a near relation in Collapsis, which is also an Antarctic deep-
sea Medusa, and which I have described in my System der Medusen (1879, p. 489, taf.
xxviii.). These two compose a special small group of deep-sea Cannostomae, which
I include provisionally as a sub-family of the Ephyridaa, but which it would be as well
to separate in future as an independent family of the Collaspidae. These two genera must
be essentially looked upon as Ephyridae, which are distinguished by their colossal size
and peculiar complications in the formation of the umbrella corona, and the coronal
intestine. The central umbrella disc, which is separated by a deep coronal furrow from
the surrounding umbrella corona, has, on the whole, the same formation as in the
Nausithoidae, especially Nauphanta. The wide, but short, quadrangular oesophagus,
cruciform in transverse section, is surrounded by eight genitalia, which in Atolla (as in
Nausicaa) are grouped in pairs, whdst in Colkqms (as in Nausithoe or Nauphanta)
they are adradially distributed at equal distances. The formation of the peripheric
umbrella corona differs entirely, as it is distinguished both by the increased number of
the marginal organs, and by special modifications of the structure. Whilst in all other
Ephyridae, Nausithoidae, as well as Palephyridae, the number of the sense clubs, tentacles,
and pairs of lobes invariably amounts to eight, in the Collaspidae it rises from sixteen to
thirty-two, and seems to vary in the same way as it does in most polynemal
NarcomedusEe. These remarkable Discomedusae are altogether so like the polynemal
Narcomedusae that at first I took them for gigantic forms of the latter. Another
peculiarity of the Collaspidae consists in the extraordinary development of their coronal
muscle. This is divided into two different, sharply defined wings, an inner or abaxial,
which is debcate and thin like a velum, and an outer or abaxial, which is dispropor-
tionately thick and divided into from sixteen to thirty-two areae. Immediately below it,
at the basis of each short tentacle, there are two thick spindle-shaped radical muscles, bke
those in the Peromedusae. The Collaspidae also resemble the Peromedusae strikingly in
the sculpture of the exumbrella, as its coronal part is divided b)r deep furrows into thick
polyhedral gelatinous pieces or pedalia. One half of these pedalia sustain the sense
clubs, the other half support the tentacles. The sense clubs and the pouches belonging
to them are small and scantily developed in Collapsis, and quite rudimentary in Atolla.
This retrograde formation of the higher organs of sense is probably a consequence of
adaptation to life in great depths of the sea. The formation of the coronal intestine is

REPORT ON THE DEEP-SEA MEDUSAE. 113

also very interesting and important, since it forms a wide coronal sinus in its proximal
part, as in the Peromedusse. This sinus, as in Nauphanta, communicates at its distal
margin with the marginal corona of pouches, and at its proximal margin by four
perradial gastral openings with the central stomach. This original arrangement has
disappeared in most of the other Discomedusse, as the four interradial septal nodes
between the gastral openings have undergone retrograde formation, and the four
perradial gastral pouches separated by them are therefore no longer present ; both these
and the coronal sinus have consequently become merged into the central stomach.

Atolla wyvittii, Hseckel (PL XXIX.).

Umbrella, quite flat, discoid, about six times as broad as high. Radius of the
central umbrella disk almost twice as large as that of the peripheric corona of the
umbrella ; disk and corona separated by a very deep coronal furrow. (Esophagus,
constricted in the middle, quadrangularly prismatic, two to three times as broad as high.
Genitalia, eight elliptical pouches, grouped in pairs, in the periphery of the oesophagus,
their perradial distances less than the interradial, 19 to 22 (16 to 32 ?) rudimentary
sense clubs, and the same number of short tentacles (half as long as the radius of the
umbrella) alternating with them. Tentacular pedalia broader and shorter than the
rhopalar pedalia. Rhopalar canals rudimentary, much narrower and shorter than the
tentacular. Marginal lobes elliptical and obtuse. Horizontal diameter of the umbrella,
58-66 mm.; vertical diameter, 10-12 mm.

Habitat. — Antarctic Ocean of the eastern and western hemispheres, in a depth of

about 2000 fathoms. The Challenger captured five specimens of this remarkable species,

which I was able to examine, preserved in spirits. The state of preservation was

unfortunately only imperfect, in spite of the tough nature of the body, the epithelia

being almost entirely wanting. All five specimens were mature females. Three of these

were taken on 3d March 1874. Lat. 53° 55' S., long 108° 35' E. Depth, 1950 fathoms.

Station 157, in the Antarctic Ocean nearly in the middle between the Kerguelen Islands

and Melbourne. The two other specimens were taken 11th February 1876. Lat.

42° 32' S., long 56° 27' W. Depth, 2040 fathoms. Station 318, South Atlantic Ocean,

St Mathias Bay, not far from the coast of Patagonia. Bottom temperature, 0*4° C. The

three Indian- Antarctic and the two Atlantic-Antarctic correspond completely in structure

and show no specific difference. The horizontal umbrella diameter in the first three (from

Station 157) amount respectively to 66, 68, and 50 mm. ; the diameter of the last two

(from Station 318) to 40 and 38 mm. The smallest specimen of the latter (38 mm.)

had only 19 tentacles and 19 pairs of lobes ; all the four other specimens had 22 tentacles

and 22 pairs of lobes. I have named this highly interesting species after Sir Wyville

Thomson, the scientific director of the Challenger expedition.

The umbrella (figs. 1-4) of Atolla wyvillii forms a circular, thick, perfectly flat disk,

(ZOOL. CHALL. EXP. PART XII. 1881.) M 15

114 THE VOYAGE OF H.M.S. CHALLENGER.

nearly six times as broad as high ; it is from 60-70 (more exactly 58-6G) mm. in
diameter, by 10-12 mm. in height. It is very firm in consistency, like cartilage. The
exumbrella (fig. 1, fig. 4, right half) is divided by a broad, very deep coronal furrow (ec)
into a central umbrella disk and a peripheric umbrella corona. The corona surrounds
the disk as a wall does a fortress, or an atoll, a circular zone of coral reefs, does the
island it encloses, from which it is separated by a ring of lagoons. The central disk
(" discus umbralis," uc) is flat, smooth, circular, and gelatinous ; its radius is nearly twice
as great as the breadth of the umbrella corona, and therefore amounts to two-thirds of
the whole radius of the umbrella. Its margin is divided by 19 to 22 radial indentations
or incisions ("sulci radiales," es) into the same number of notches or teeth. These
notches of the disk are quadrangular, 5 mm. broad and nearly as long ; their height at
the external perpendicular fall into the "mural ditch" amounts to 8 mm. The teeth lie
in the same radii as the rhopalar pedalia (uo) ; they alternate with the tentacular pedalia
(>it), which correspond to the radial furrows between the teeth.

The exumbral coronal furrow ("fossa circularis," ec) in Atolla is so deep, that in its
fundus the central umbrella disk is only connected with the peripheric umbrella corona
by a very thin gelatinous ring (fig. 4, ec). Its depth amounts to 6-7 mm. ; its greatest
breadth (in the lower third) to 4 mm. It resembles a circle of lagoons, which separates
the central island from the surrounding atoll reef, or the deep ditch which separates the
enclosed fortification from its circular wall. But as the teeth of the umbrella disk, pro-
ject towards the outside with their upper edge over the mural ditch, like overhanging cliffs,
the upper, cleft- shaped passage into the coronal furrow appears only 1-2 mm. broad.

The peripheric corona of the umbrella (" corona umbralis ") is half as broad as the
radius of the central umbrella disk, and is composed of three different zones : an inner
zone of tentacular pedalia (ut), a middle zone of rhopalar pedalia (uo), and an outer zone
of marginal lobes (I). The inner zone consists of from 19 to 22 gelatinous sockels of the
tentacles ("pedalia tentacularia," ut). They are thick, almost dice-shaped gelatinous
pieces, which lie close together and are only separated by shallow radial furrows. The
lateral length of this gelatinous die amounts to nearly 6 mm. Each pedalium forms the
basis or socket of a tentacle, which springs from its outer surface. On closer con-
sideration we perceive the following conditions of form : — The upper (aboral) surface
(fig. 1, ut) is smooth, arched rather convexly and hexagonal ; of the six side lines of this
hexagon, the inner are contiguous to the coronal furrow, and lies opposite the radial furrow
(es) between each two adjacent teeth of the central disk (er). The two inner lateral
lines of the hexagon are contiguous to the corresponding lateral lines of the two adjacent,
whdst of the three external lines, the middle touches the tentacle basis, and the two
outer lateral lie opposite the two adjacent rhopalar pedalia (uo). The lower (oral)
surface of the tentacular pedalia has the form of a parallel trapeze, and forms the upper
wall of a tentacular pouch. Their inverted lateral surfaces are separated by a radial

KEPOET ON THE DEEP-SEA MEDUS-E. 115

furrow, which corresponds to the middle line of the disk teeth, and of the rhopalar
pedalia. Their axial surface forms the outer, almost perpendicular or only slightly
overhanging wall of the coronal furrow, whilst their abaxial surface serves for the insertion
of the proximal tentacle bulb.

The gelatinous sockels of the sense clubs ("pedalia rhopalaria ") alternate regularly
with the gelatinous sockels of the tentacles, and form the second middle zone of the
umbrella corona (figs. 1, 4, no). They are nearly as large as the tentacular pedalia,
somewhat longer but not so thick, and are inserted with their proximal part between the
distal sides of the latter. Their upper aboral surface is almost pentagonal, and depressed
(fig. 1, uo). Their lateral margins are separated by a broad interspace which is filled
by the bulb of the tentacle. Its truncated distal margin bears a pair of thin marginal
lobes and the rudiment of a sense club in the incision between the pair.

The marginal lobes (" lobi marginales," fig. 4, I) form the third or outer zone of
the umbrella corona. They amount in number from 38 to 44, as each marginal
lobe is inserted between a tentacle and a rhopalium. Their shape is a longish round,
2-3 mm. broad, 5-6 mm. long. The proximal third of each marginal lobe consists
of a thick semi -oval gelatinous part, which is merely the distal bifurcation of a rhopalar
pedalium. The middle and distal third of the marginal lobe is formed by a very
thin membranous, folded marginal border (" patagium," Ip). This was invariably torn
and badly preserved.

The gelatinous disk (wo/) of Atolla is thick and firm (fig. 4, left half, in vertical
meridian sections). It has the consistency of a tolerably firm fibrous cartilage. In the
central umbrella disk (uc) its thickness amounts to 5 mm., even to 10 mm. at the
thickest part at the marginal teeth (er) ; whilst immediately outside these, at the thinnest
part of the coronal furrow (ec'), it is only h mm. The gelatinous substance of the pedalia
is from 4-7 mm. thick.

The umbrella margin of Atolla includes all the parts lying outside the coronal muscle,
and is therefore composed of from nineteen to twenty-two tentacles, the same number of
sense clubs alternating with them and twice the number of marginal lobes, inserted
between the former and the latter. The rhopalia at the distal margin of the rhopalar
pedalia are, however, so small and the two marginal lobes at the distal end of the pedalia,
which enclose the rhopalia, are so closely united, that at a superficial glance it looks as if
the umbrella margin was merely composed of alternate tentacles and rhopalar pedalia.
Closer investigation and comparison with the more completely developed umbrella
margin of the closely allied Collapsis (System, pi. xxvii.) shows, that the umbrella
margin is essentially composed like that of the latter, only that the sense clubs and their
pouches have undergone much greater retrograde formation in Atolla.

The tentacles (figs. 1-4, t) are very weak, hardly half as long as the radius of the
umbrella, they are shaped like an awl and finely pointed towards the thin end.

11(5 THE VOYAGE OF H.M.S. CHALLENGER.

Their basal part is strongly thickened and forms a conical tentacle bulb, which fills the
interspace between each two rhopalar pedalia, and is inserted with a broader base at the
distal side of the tentacular pedalium. This basal part is also hollow and contains the
csecal end of the thin tentacular canal, whilst the distal part of the tentacles is solid (as
in Nauphanta). A strong longitudinal muscle runs both on the upper and the lower
surface of the tentacle. The upper or external, abaxial tentacle muscle (fig. 4, mt') is
the shorter and weaker, only occupies the proximal third of the tentacle, and passes
to the outer margin of the upper surface of the tentacular pedalium. The lower or
internal axial tentacle muscle (fig. 4, mt") is longer and stronger, runs along the whole
length of the tentacle, and passes with two very strong, fusiform radial muscles (mk,
fig. 3, above, to the right) to the umbral surface of the tentacular coronal pouch (as in
PeriphyUa, PL XXIL).

The sense clubs (" rhopalia," or) in Atolla are quite rudimentary and more slightly
developed than in any other Discomedusae hitherto examined, in fact this might be easily
overlooked, as their obscure rudiments lie hidden at the distal margin of the rhopalar
pedalia, between the basis of the two marginal lobes of a pair. It was only with consider-
able trouble that I succeeded in determining their existence ; they alternate regularly
with the tentacles, so that their number also amounts from nineteen to twenty-two.
Their anatomic nature could unfortunately not be found out on account of their small size
and the bad preservation of the umbrella margin in all five specimens ; but as the sense
clubs are indubitably in the same position and better developed in the closely allied
Colkqms (System, pi. xxviii. figs. 3, 4), there can be no doubt as to the significance of
the small rudiments in Atolla. We have probably to do here, as in many other
deep-sea animals, with a phylogenetic retrograde formation of this organ of sense.

The subumbrella (figs. 2, 3, 4) is divided in the same way as the exumbrella, by the
deeply incised coronal furrow, into two separate principal arese, which are only connected
by the thin gelatinous ring (ec) at the bottom of the coronal furrow. The central area
of the subumbrella is therefore the same size as the central disk of the umbrella ; it is
formed by the gastrogenital membrane, which reaches as far as the distal margin of the
coronal sinus, and contains the stomach in its central part and the corona of eight
genitalia (s) and their alternating deltoid muscles (md) in the peripheric part. The
deltoid muscles are narrow and slightly developed, especially the four perradial (md')
whilst the four interradial appear to be considerably broader (md"). All the eight
deltoid muscles in Atolla are triangular only in the distal half, and rectangular in the
proximal half (between the genitaba) ; the interradial muscles are inserted on the base
line of the cathammal arese (kt), the perradial at the distal margin of the gastral
openings (go).

The coronal area of the subumbrella begins at the distal margin of the genitalia, and
is separated from central area by the very thin ring of the gelatinous disk, which

REPORT ON THE DEEP-SEA MEDUSAE. 117

forms the bottom of the deep coronal furrow. The coronal area of the subumbrella
in Atolla consequently corresponds exactly to the umbrella corona of the exumbrella,
and is likewise divided into three zones : the inner zone of the internal coronal
muscle, the middle zone of the external coronal muscle, and the outer zone of the
marginal lobes. The broad, strongly-developed coronal muscle (" musculus coronaris ")
consists in Atolla, as in Collapsis (System, taf. xxviii.), of two separate sharply-
defined halves. The inner coronal muscle (" musculus coronaris internus," mc)
is 5 mm. broad, thin and delicate, and extends like a veil over the inner zone
of the coronal area of the subumbrella. It leaves the proximal third of this zone free,
as it does not extend as far as the coronal furrow ; it forms at the same time the
proximal third of the coronal pouches, which occupy the greater part of this zone. The
outer coronal muscle (" musculus coronaris externus," mc") is only 4 mm. broad but
extremely thick ; like the outer muscle it consists entirely of circular muscular fibres ;
these are accumulated in many layers one above the other in such a way that they repre-
sent a band-shaped circular muscle, 2 mm. thick. This extremely strong fleshy mass
belongs to the most powerful muscular formations hitherto observed in the Medusae
(comp. the transverse section, fig. 4, left, mc", and figs. 7, 8, mc"). The 19 to 22
deep radial furrows of the subumbral under surface, which correspond to the tentacles,
divide the outer coronal muscle into the same number of sections (figs. 2, 3). Whilst
the sharply-defined external coronal muscle forms the middle zone of the coronal area
of the subumbrella, its external zone occupies the corona of marginal lobes ; at the
subumbral side of each lobe we find a weaker longitudinal muscle, which radiates
into the thin membranous and folded marginal border, the patagium.

The umbrella cavity in Atolla is very small, corresponding with the flatness of the
disk. As the wide oesophagus reaches to its opening and fills its axial space, the
umbrella cavity actually merely consists of the narrow, circular, hollow space, between the
external wall of the oesophagus and the corona of genitalia. Between the four perradial
mesenteric folds of the stomach (ivr) it is depressed in the form of four conical niches
projecting inwards, which may be considered interradial funnel cavities, although only
of small extent and depth (figs. 1,3, ii).

The gastrovascular system (figs. 3-6) of Atolla is closely allied in many and important
respects to that of the foregoing Nawphant a, but still shows several peculiarities which
remind us partly of the Tesseronise (Peromedusae), partly of the Semostomas. Of the two
principal sections, the axial principal intestine is very simply formed, in the shape of a
quadrangular, depending oesophagus, whilst the peripheric coronal intestine shows very
complicated formations, and extends in the form of a horizontal corona of pouches,
communicating with the axial intestine by four perradial gastral openings.

The central principal intestine (" gaster principalis," figs. 3-6, g) forms a short, wide
quadrangular oesophagus, which hangs freely from the central part of the umbrella disk,

118 THE VOYAGE OF H.M.S. CHALLENGER.

and fills the greater part of the umbrella cavity, its oral margin (am) reaches to the
opening of the latter. Its typical form is a regular quadrangle prism, whose lower
surface occupies the simple cmadrate oral opening, whilst the upper surface is formed by
the endodermal surface of the central gelatinous umbrella. Both these two surfaces and
each horizontal transverse section of the oesophagus describe a regular cross, as four
perradial cross limbs project centrifugally the whole length, whilst the four interradial
oral columns (ac) project inwards centripetally. The largest horizontal diameter of the
gastral hollow space (in the perradia) amounts, both at the base of the stomach and at
the oral opening, to from 22-34 mm., whilst the smallest diameter (in the interradia)
amounts to only half as much, 11-12 mm. The entire height or length of the oeso-
phagus is still less, amounting from the base to oral margin, only from 8-10 mm.

The gelatinous fulcral plate (zw) is strongly thickened in the upper half of the
subumbral gastral wall, and forms several depressed elevations (fig. 5, gw) on its inner
surface. It is very delicate and thin, however, in the lower oral half. The circular
muscular layer of the oesophagus is also only slightly developed. As in Nauphanta
(PI. XXVIII. fig. 14), there is a circular constriction in the middle of its length,
which divides the oesophagus into two chambers, shaped like truncated pyramids, which
are connected by their narrow bases. We may, perhaps, compare this circular stricture, as
in Nauphanta, with the palatine door of the Tesseronise, in which case the lower chamber
(which widens below, towards the mouth) must be regarded as the buccal stomach or oeso-
phagus, and the upper chamber (which widens above, towards the bottom of the stomach)
as the central stomach fused with the basal stomach. In transverse section, through the
circular stricture or palatine opening (fig. 6), the largest (perradial) diameter of its
cruciform lumen only measures 15 mm., the smallest (interradial) only 6 mm. Below the
circular stricture the thin perradial walls of the buccal stomach project, inflated to the out-
side, and form buccal pouches (fig. 3, bb) which are separated by interradial buccal columns
projecting inwards (wc), as in Periplujlla (Pis. XVIII.-XX.), but not so strongly developed.
Above the palatine opening,